[
    {
        "videos": [
            "videos/eval_0/0.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Black man in a short-sleeve shirt stands at a kitchen stove. He holds a box of dry pasta in one hand and pours the pasta into a pot of boiling water. Using a wooden spoon, he gently presses the pasta down to ensure it is fully submerged. The background shows kitchen counters and utensils. The camera remains steady, focusing on the man’s hands and the pot..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage Symbols. [Motion & Physical Consistency]: Good. [Prompt Alignment]: The subject's face is not visible, unable to identify gender.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/1.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black man with a beard walks through an aisle of a brightly lit toy store, surrounded by colorful shelves. He pauses in front of a shelf displaying puzzle sets, picks up a puzzle set in both hands, examines the pieces closely, and smiles at the memories of his own childhood. The camera remains steady, capturing his actions and the vibrant store setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: His right index finger passes through the object he's holding unnaturally. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/2.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The boot slowly rotates clockwise, showcasing its rugged sole and detailed leather texture against a blurred background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The video has a garbage caption. [Motion & Physical Consistency]: Good [Prompt Alignment]: The boots does not rotates clockwise.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/3.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman lifts a colorful shopping bag toward the camera with a wide grin, revealing neatly packed beauty and skincare products inside..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: The overall movement, the deformation, and the cause-and-effect involving the shopping bag all seem unnatural, she moves it, yet the beauty and skincare products inside shows no movement. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/4.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A mature Middle Eastern man wearing a traditional red and white checkered keffiyeh and a white thobe sits inside a softly lit car. He faces the camera, maintaining steady eye contact while speaking in a calm, gentle manner. His expression is relaxed, reflecting a sense of relief. The car window reveals a subtle view of the outside. The camera remains fixed, capturing his natural demeanor in the quiet, personal setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The texture of the keffiyeh is flickering. [Motion & Physical Consistency]: Deformed teeth [Prompt Alignment]: He is not maintaining steady eye contact.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/5.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged man in a navy shirt stands in his home office. He uses both hands to lift his black desk to a standing position, then picks up a pen and begins writing in a notebook on the desk. The background shows bookshelves and a window with daylight streaming in. The camera remains steady, capturing his actions and the organized office setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: The black desk seemed deformed when he raised it. He did let go of his left hand but the table still floats and moves. The writings on the paper are inaccurate to the movement of the pen. The pen came out of nowhere. [Prompt Alignment]: Good.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/6.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Caucasian woman with a ponytail, wearing a light athletic top, sits at a kitchen table. She holds a glass of iced tea in one hand and uses a straw to stir the drink, watching the liquid swirl. Sunlight streams through a nearby window, and a water bottle rests on the table. The camera remains steady, capturing her relaxed posture and the refreshing setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The texture of her athletic top is flickering. [Motion & Physical Consistency]: Good [Prompt Alignment]: The window is not visible we cannot assume that it comes from the window.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/7.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The hand gently holds the black device, with a subtle and smooth movement that highlights its details against the car dashboard and steering wheel in the background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/8.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A medium shot underwater shows two women with long, light-colored dresses and flowing hair. Both women float gently in the water, with their dresses and hair drifting naturally. The woman in the foreground steps forward with light arm gestures, while the woman in the background spins slowly, causing her dress to swirl around her. The camera smoothly shifts angles several times, capturing their synchronized movements and the shimmering water..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: texture issues on the water [Motion & Physical Consistency]: woman in the background is distorted [Prompt Alignment]: The woman in the foreground DID NOT step forward with light arm gestures.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/9.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A 16-year-old woman with pale lips and delicate facial features lies motionless on a wooden coffin bed in a dimly lit indoor mourning hall. She wears traditional white mourning robes and has her long black hair scattered across her shoulders. The background features soft candlelight, incense smoke, and white funeral lanterns. The camera captures her from a low angle, highlighting her stillness and the ghostly atmosphere. The shot is framed in a 9:16 vertical format, with subtle shadows and a cold color grading, resembling a scene from Chinese ancient drama..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: She doesn't look 16. There are no white funeral lanterns visible in the video.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: No. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/10.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged man with tan skin and salt-and-pepper hair walks into a minimalist bedroom, pulling a navy rolling suitcase with glossy finishes. The suitcase glides effortlessly across the hardwood floor as his movements exude calm precision. He stops near a window, bathed in warm, golden light, and turns toward the camera, offering a polite nod while gripping the handle firmly. The final frame focuses on the suitcase, showing its sturdy wheels and textured exterior, with soft lighting highlighting the cozy atmosphere of the room..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Distorted left hand and moves  unnaturally. Stabilize to make it look natural. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/11.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black man in a bright orange hoodie pushes a small gray car along a quiet countryside road in the rain. The headlights of the car reflect off puddles on the wet pavement, and the soft, diffused evening light creates a calm atmosphere. The camera remains steady, capturing the man’s effort and the rain-soaked rural background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: texture issue on wet roads/puddles [Motion & Physical Consistency]: good [Prompt Alignment]: good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/12.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A woman with medium skin tone and curly hair stands in her upscale kitchen, wearing fitted yoga leggings and a matching sports top. She glances around the bright space, gently stretching her arms above her head before lowering them to her sides. The background features marble countertops and stainless steel appliances. The camera remains steady, capturing her relaxed posture and the elegant kitchen setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Deformed hand [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/13.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A dark-skinned teenager with braided hair sits cross-legged on a blanket in an open grassy field at sunset. The subject holds a pink wool flower between the fingers of the left hand and uses a felting needle in the right hand to shape and adjust the petals. The subject repeatedly pokes the needle into the flower, pressing and pulling the wool to refine its form. The camera remains steady, capturing the subject’s hands and the textured wool against the golden evening sky and moving grass..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Good. [Prompt Alignment]: She didn't hold a pink wool flower between the fingers of her left hand.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/14.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A realistic handbag rests on a marble table. A hand slices the handbag down the middle, revealing layers of chocolate cake and frosting inside. The camera remains steady, clearly showing the cross-section of the cake within the handbag..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Good. [Prompt Alignment]: Good.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/15.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A metal spatula glides smoothly through a thick layer of white whipped cream, creating clean, even trails in the cream. The scene is set against a vibrant, colorful background. The camera remains steady, focusing closely on the spatula’s motion and the texture of the whipped cream..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The mark of the object or the trail is not accurate and not consistent. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/16.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A group of adults of various races and genders follow a Zumba instructor’s high-energy dance routine in an industrial-style dance hall. The participants move in sync, stepping and swinging their arms as they mirror the instructor’s motions. The background features exposed brick walls, wooden floors, and large garage-style windows, creating a warm, urban atmosphere. The camera captures the group’s coordinated movements and the dynamic setting in a wide, steady shot..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: No [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: No - It looks like it was generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/17.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Caucasian man applies a subtle lipstick shade, his lips looking naturally enhanced..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Cause-effect fails as the marks of lipstick are inaccurate. [Prompt Alignment]: The subject is not Caucasian.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/18.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: In a futuristic underwater dome surrounded by marine life, a young East Asian man in a sleek jumpsuit studies a suspended sphere of golden, glimmering water that floats mid-air. He delicately scoops the liquid using a transparent glove, marveling at the sparkles dancing like liquid light as fish and corals glow softly in the background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Unrealistic bubbles flowing out of the sphere glass. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: He didn't scoop the liquid as his hand didn't even touch the sphere glass..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/19.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged Black man with short hair stands in a well-lit bathroom. He uses his fingertips to massage a rich face cream into his forehead and temples in gentle, circular motions. The background shows a bathroom mirror and tiled walls. The camera remains steady, focusing on his face and hands as he works the cream into his skin..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: entire face was blurry and pixelated. [Motion & Physical Consistency]: good [Prompt Alignment]: good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/20.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly South Asian man wearing a beige sweater adjusts his glasses, holding a green dart in his right hand as he pauses to concentrate on the bullseye on a dartboard hanging in a garage cluttered with tools and bicycles..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is a subject-object interaction issue as the dart appears passing through the man's eyeglass. There is also a cause effect issue as the dart in the dart board moves on its own. [Prompt Alignment]: The man is not South Asian. Tools are not visible in the video.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: No. Object Existence: No. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/21.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Indonesian man in his 20s stands beneath a concrete bridge as heavy rain pours down around him. He wears a dark hoodie and jeans, holding a lit cigarette between his fingers, his tired eyes gazing ahead. Raindrops splash on the wet ground behind him, creating shimmering reflections on the concrete. The scene is captured in warm, moody tones with cinematic film style lighting. The camera remains steady, highlighting the realistic texture of his skin and hair as rain falls in the background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The cause-effect of the lighting flame is unnatural. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/22.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Asian woman with an athletic build sits confidently in a modern esports streaming studio. She wears a black hoodie and loose-fitting pants. The background shows RGB lighting, gaming monitors, and subtle esports memorabilia, creating a professional atmosphere. She faces the camera with steady eye contact, her expression composed and confident. Her movements are natural and minimal as she speaks, emphasizing an engaging and sincere conversation style. The camera remains fixed, capturing her focused demeanor and the immersive studio setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: pixelated hair. [Motion & Physical Consistency]: good [Prompt Alignment]: not able to confirm if subject is wearing loose-fitting pants.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/23.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Caucasian man uses a hairbrush to style his short hair, ensuring it’s neat and perfectly shaped..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: NULL [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/24.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black man stands in front of a plain, light-colored wall. He holds a fry in his right hand and brings it toward his mouth to eat. In front of him on a table is a large pile of food, including a burger and fried items. The camera remains steady, capturing the man’s action and the food on the table..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is a cause effect issue as the bite mark was not accurate. [Prompt Alignment]: He can't be determined due to lack of view. The table was not visible.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/25.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young South Asian woman snaps a picture of a sunset with her phone, her eyes focused on the screen as the light plays off her outfit..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Good [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/26.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A tall white adult man with long hair stands on a mountainside beside a bubbling stream. He holds a dripping tie-dyed scarf in both hands, displaying its vibrant colors—purple, pink, and orange—against the backdrop of tall pines. Water flows nearby, and the camera remains steady, capturing the man, the scarf, and the forested landscape..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/27.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Hispanic man applies moisturizing cream to his face in a minimalist bathroom. The camera shows his fingers rubbing the cream in, focusing on the smooth texture and how it hydrates his skin, giving it a healthy glow..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Cause-effect fails as the cream wasn't spread accurately. [Prompt Alignment]: Good.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/28.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: From inside a shopping bag, the camera shows a Southeast Asian woman reaching in. She grabs a pastel-colored sweater, and as she lifts it out, the soft fabric brushes against the edge of the bag. The background is softly lit, with glimpses of a clothing store’s shelves and racks visible through the bag opening. The camera remains fixed inside the bag, focusing on her hand and the sweater as they move..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Interaction with the sweater as it deforms is not in accordance to physical law. Cause-effect failure as it changes its color and another sweater suddenly appears. [Prompt Alignment]: The subject is not entirely visible.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/29.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The subject, a Hispanic man, eats a loaded hotdog with relish and onions, the toppings adding a tangy contrast to the savory sausage..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Cause-effect failure as the food seems to have no bite at all even if he has already eaten part of it and it directly entered his mouth. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/30.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman with curly hair sits on a bench in a sunlit park. She holds a sketchbook on her lap and uses her right hand to carefully sketch a sleek, modern watch, her gaze fixed on the page. Her left hand steadies the sketchbook. The background shows green grass, leafy trees, and people relaxing in the distance under bright daylight. The camera remains steady, capturing her focused expression and precise hand movements..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The cause and effect of the drawing is unnatural. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/31.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Black man applies a drop of serum directly onto his nose, where textured pores are visible, the formula spreading across the skin under soft lighting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the cause-effect as it was seen that the liquid was drop to the nose but it was seen that it sudden form to a straight formation to the nose and it was not interacts with the subject and it was appearing unrealistic. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: There is an issue with the cause-effect as it was seen that the liquid was drop to the nose but it was seen that it sudden form to a straight formation to the nose and it was not interacts with the subject and it was appearing unrealistic. .\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/32.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A 45-year-old person with light skin and short brown hair stands on tiptoes in a bright, American middle-class kitchen with stainless steel appliances and wooden cabinets, reaching for a glass jar on a high shelf. As they stretch upward, they suddenly freeze, bend forward, and grip their lower back with their fingers pressing into the cotton T-shirt. Their face contorts in pain: eyebrows knit, teeth biting their lip, eyes squeezed shut, and sweat visible on their forehead. A small, vivid red glow appears at their right lumbar area. The glow trembles and intensifies for two seconds, pulsing in sync with their flinch to indicate a sharp burst of lower back pain. The camera remains steady, clearly showing the kitchen background and the subject’s reaction..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Deformed hand [Prompt Alignment]: He is not a 45-year-old person\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/33.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged Black woman in a dark insulated jacket and snow pants stands on a snowy mountain slope dotted with evergreen trees. She launches forward, gripping the handles of a dark green paraglider that lifts above her as she takes off. As she glides just above the glistening white snow, she smiles and briefly glances at the camera. The background shows snow-covered evergreens under bright daylight. The camera follows her smooth launch and flight from the side..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the cause-effect as the marks of the snow was inconsistent as the board was not touches on the the ground but it keeps bursting with a snow and making it appears unnatural. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/34.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman dressed in a business suit sits outside at a café with a laptop and coffee; the scene freezes on her shocked expression as her chair tips backward on uneven cobblestones..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: The subject's movement is stiff and looks robotic. There is deformation in the subject's eyes. Stabilize to make it look normal. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/35.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The shot moves along a line of people waiting at a food truck, revealing a diverse mix: a Latina woman in sunglasses, a white guy with a backpack, a Middle Eastern teen holding a drink, and a Black man wearing a graphic tee and sneakers..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the subject interaction as the hand of the subject was seen having a deformation. [Prompt Alignment]: There is no white guy in the backpack.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/36.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Middle Eastern man wearing yellow swim trunks stands at the edge of a rocky cliff in a tropical landscape. He leaps forward and dives headfirst into a crystal-clear river below. The camera follows his freefall as he descends, capturing the moment he enters the water with a clean splash. Lush green trees and plants surround the river in bright daylight..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: No [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: No - It looks like it was generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/37.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The camera slowly and steadily zooms in on the APEX HAIR container, highlighting the applicator stone inside, while the digital timer above remains visible..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage symbols [Motion & Physical Consistency]: The cause-effect of the floating product is unnatural. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/38.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young East Asian man stands at a table covered with a large pile of crumpled banknotes. He uses both hands to pick up a single banknote from the top of the pile, then holds it up to the camera to display its front and back. He repeats this motion, picking up another banknote from the pile and holding it up for display. The background shows a plain tabletop and soft indoor lighting. The camera remains steady, focusing on his hands and the banknotes..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage symbol [Motion & Physical Consistency]: Deformed face [Prompt Alignment]: Cannot determine if it's an East Asian / a man. The man did not pick up another banknote from the pile and held it up for display\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/39.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An older Middle Eastern woman stands in a study room, wearing a traditional headscarf that fully covers her hair and neck. She has beautiful, well-defined facial features. She is dressed in a royal blue linen robe, featuring light beige floral embroidery with silver thread along the collar and dark embroidered seams at the cuffs, paired with formal beige pants decorated with dark patterns. Facing the camera, she speaks while making simple hand gestures. As she talks, she smoothly sits down. The background shows bookshelves and a desk under soft indoor lighting. The camera remains steady, capturing her movements and attire..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The texture of the embroidered design is flickering. [Motion & Physical Consistency]: Good. [Prompt Alignment]: Good.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/40.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: Inside a cozy bathroom with soft yellow lighting, an elderly South Asian man uses a wax applicator to spread golden honey-textured wax onto his own back, the steam from a hot shower in the background slightly blurring the mirror..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: NULL [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/41.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: Inside a futuristic pastel-toned meditation pod filled with soft pink lighting and holographic projections of serene landscapes, a middle-aged South Asian man with deep brown skin and short salt-and-pepper hair wears a pale pink sheet mask. His laughter shines through as his nimble fingers wiggle playfully beneath his chin, reflecting the glowing virtual petals drifting in the background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Unnatural movement of his body. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: I think it's generated by AI\n.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/42.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Hispanic man in his 30s with short dark hair stands in a sunlit outdoor setting. He wears a light blue button-down shirt and balances classic aviator sunglasses on the tip of his nose. He looks over the top of the sunglasses at the camera with a relaxed smile. The reflective lenses catch the light. The camera remains steady, capturing his casual posture and the bright background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: He did not looks over the top of the sunglasses .\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/43.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged man with tan skin and greying hair carefully lines up a golf putt on a sunlit putting green. He grips a sleek, silver putter, the club gleaming softly as rays of sunlight glance off its polished surface. The camera zooms in as he gently swings, sending the ball rolling toward a flagstick in the distance surrounded by lush grass..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Good.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/44.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Hispanic woman in her late twenties with long dark hair sits at a kitchen table. She lifts a glass of iced coffee to her lips and takes a sip, causing the ice cubes to clink against the glass. The background shows a bright kitchen with sunlight streaming in. The camera remains steady, focusing on her action and the drink..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The cause and effect of the coffee seems unnatural. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/45.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Southeast Asian man stands in front of a bathroom mirror, wearing a sleeveless shirt. He dispenses lotion onto his hands and applies it to his shoulders and chest, using firm, circular motions to massage it into his skin. The bathroom background includes a sink, tiled walls, and soft overhead lighting. The camera remains steady, focusing on his upper body and reflection as he completes the action..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Cause-effect failure as the cream didn't spread accurately. [Prompt Alignment]: He didn't apply lotion to his chest.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/46.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The subject, a Hispanic woman, applies mousse to her hair, ensuring even distribution as she styles her locks..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: His hand movements are unnatural. Interaction fails as her right hand deforms unnaturally. The mousse in her hair is unnatural. [Prompt Alignment]: She isn't Hispanic.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/47.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A single French fry, resembling a potato chip, is seen floating on the surface of an ocean with waves. The camera quickly pulls back to reveal a wide view from above, showing the fry riding a wave and moving across the water. As the camera moves upward, it becomes clear that the fry is actually a chip and is being pushed by the wind, performing a windsurfing maneuver. The background features a large number of chips scattered on the water, some partially submerged and others floating. The camera remains in a steady overhead shot throughout the sequence..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Flickering issues. [Motion & Physical Consistency]: Good. [Prompt Alignment]: It didn't show the fry riding a wave and moving across the water.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/48.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A white man in his 30s, wearing a gray hoodie, stands at a kitchen counter. He holds a can of cleaning foam in his right hand and sprays a layer of foam directly onto visible coffee rings and crumbs on the countertop. The foam spreads and covers the stained areas. The background shows kitchen cabinets and a window with daylight. The camera remains steady, focusing on the man’s action and the countertop..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is a cause effect issue as the cleaning foam was inconsistently appearing and the crumbs are increasing. [Prompt Alignment]: There is no coffee rings visible.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/49.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged South Asian man in a short-sleeve shirt stands at a stove in a bright kitchen. He uses a metal spatula to toss chopped herbs, garlic, and green chilies from a hot wok, causing them to fly upward in slow motion. Droplets of oil catch the light as the ingredients rise and fall. The camera remains steady, focusing on the wok and the airborne ingredients against a tiled kitchen background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: texture issues in the ingredients [Motion & Physical Consistency]: good [Prompt Alignment]: unable to determine if man is middle aged and south asian\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/50.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A South Asian woman stands in a well-lit bathroom. She dispenses cream cleanser onto her fingertips and gently rubs it over her face in circular motions to cleanse her skin. The camera remains steady, showing her face and upper body against a background of tiled walls and a bathroom mirror..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: No [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: No - It looks like it was generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/51.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged East Asian woman sits in the passenger seat of a compact car illuminated by soft light. Her nails shimmer faintly as she gestures and talks sharply, emphasizing her intonation. Through the car’s window, a dark night sky is visible in the background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Blurry foreground [Motion & Physical Consistency]: The body movement is not natural. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: The video is unrealistic..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/52.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young woman with light skin and shoulder-length brown hair stands in a well-lit living room. She holds a rectangular canvas with a cross design in both hands, displaying its texture and color. She then places the canvas on a wall-mounted bracket, gently adjusting it to hang straight. The background shows a sofa and decorative plants. The camera remains steady, focusing on her actions and the canvas..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: good [Motion & Physical Consistency]: hand deformity [Prompt Alignment]: The canvas's connector was not mounted onto the bracket and canvas looks suspended in air\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/53.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A close-up captures a white woman’s lips slightly parted as she swipes on a coral glossy lipstick, creating a smooth, glass-like finish..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The movement of the woman while applying the lipstick is not natural, as it appears aimless. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: The movement of the woman while applying the lipstick is not natural, as it appears aimless. I believe the video is AI-generated because of this discrepancy..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/54.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Japanese woman with light skin and neatly manicured nails stands in a well-lit bathroom. She holds a small tube of cream in her right hand and squeezes a small amount onto her left wrist. Using her right hand, she gently spreads the cream across her wrist in a circular motion. The background shows a bathroom counter and mirror. The camera remains steady, focusing on her hands and the application process..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The texture of the finger is flickering. [Motion & Physical Consistency]: The deformed hand and the cause and effect of spreading seem unnatural. [Prompt Alignment]: The subject's face in the video is not visible, so we cannot identify whether she is a Japanese woman.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: No. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/55.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A subject sits at a kitchen counter and holds a blue, sauce-covered food item close to their mouth, preparing to eat.First, the subject holds a large, round blue food item covered in a dark brown sauce and sprinkled with a granular seasoning. The subject brings the food item toward their mouth. Then, the subject takes a bite from the food item while holding a large black plate underneath, which is covered with remnants of the same sauce and seasoning. The subject continues to hold the food item near their mouth in subsequent frames..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: There is an issue with the cause-effect as the food was seen having a saucy substance but upon eating no visible marks was seen on his face and it appears a liquid sauce was dipping unrealistic. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/56.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A mature, handsome African man stands facing the camera at a press conference. He wears a blue-gray houndstooth blazer over a light blue pinstriped shirt, with dark khaki windowpane tapered trousers. He has a confident CEO demeanor. Behind him, a large screen dominates the background. He speaks directly to the camera, occasionally glancing around the room, as the steady camera captures his poised stance and professional setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The texture of the blazer is flickering. [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/57.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black man wearing a winter jacket and boots walks steadily through fresh snow. His boots press into the white powder with each step. The background shows a snowy landscape with bare trees and overcast sky. The camera remains steady, capturing his movement and the snowy surroundings..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: No [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: No - It looks like it was generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/58.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A teenage boy with light freckles and sandy brown hair kneels beside a small tent in a moonlit forest clearing. He holds a smartphone in his left hand, using the screen’s reflection as a mirror. With his right hand, he carefully brings a contact lens toward his right eye, focusing intently on the screen. The background shows tall trees and soft moonlight filtering through the branches. The camera remains steady, capturing his concentrated expression and the quiet nighttime setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The movement is aimless and looks weird. There is deformity in the subject's hands. [Prompt Alignment]: The subject did not bring contact lens toward his right eye. There are no contact lenses visible.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/59.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An adult crouches low to the ground in an industrial workshop with exposed brick walls and metal scaffolding. Wearing protective gloves and a welding mask, they hold a welding torch with both hands and position it over a metal rod, creating bright sparks and a concentrated arc of light. Several welding rods and metal pieces are scattered on the ground nearby. The camera remains steady, capturing the subject’s focused welding motion and the surrounding workspace..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Good [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/60.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A medium shot shows an adult Malay man with short hair and casual clothing standing at a busy Singapore hawker centre. He smiles and stands in front of a small table displaying a neatly plated nasi lemak. He gestures toward the different ingredients on the plate—rice, fried chicken, egg, peanuts, and sambal—and briefly points to each item. He then adjusts the plate for the camera, picks up a spoon, and takes a small bite. The camera slowly pans right, capturing the lively background and his actions..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage symbols [Motion & Physical Consistency]: Deformed background and there is no marks of the food in his finger as he touches it. [Prompt Alignment]: The subject's face is not visible, it does not look like a sambal and he did not picks up a spoon, and takes a small bite.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: No. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/61.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman with curly hair stands on a brightly lit baseball field at sunset, gripping a wooden baseball bat. She swings the bat forcefully and hits a spinning white baseball, sending it into the air. The background shows a faintly cheering crowd and a golden sky. The camera remains steady, capturing her swing and the field behind her..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: No [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: No - It looks like it was generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/62.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: In a serene park, a Black woman turns the cap to open a plastic bottle of water. The camera focuses on her hand as she opens the bottle, showcasing the clean, fresh water inside as she prepares to take a sip..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Cause-effect failure as the cap floats. [Prompt Alignment]: The subject is not entirely visible.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/63.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: In a cozy living room, a South Asian woman with long hair applies a brightening skin mask to her face. The camera captures her hands as she gently massages the mask in, leaving her complexion glowing and refreshed..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: There is deformity on the hands, making it unnatural. Cause-effect failure as the skin mask is not consistent as she massages it. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: She's not South Asian..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/64.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly East Asian man in a crisp white shirt stands at a wooden table in a warmly lit kitchen. He holds a pastel pink spatula in his right hand and uses it to spread a creamy, pearl-like substance across the table’s surface in smooth, fluid motions. The background shows blurred herbs and spice jars, adding rustic charm. The camera hovers directly above, tightly framing the spatula’s movements and the glossy shine of the cream as he creates gentle, even patterns. The lighting highlights the material’s sheen and the careful motion..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: The face is not visible. The camera did not hover above.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/65.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A South Asian woman in athletic clothing stands in a well-lit gym with kettlebells and exercise mats in the background. She grips a kettlebell with both hands and swings it forward and upward to chest height, then lets it swing back between her legs, repeating the motion smoothly. The camera remains steady, capturing her controlled form and the gym setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Yes [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: No - It looks like it was generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/66.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman in a towel squeezes lotion into her hand and smooths it over her arms in long, even strokes, the light catching the moisture on her skin..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The movement of the woman while applying the lotion is not natural as her hand was deforming. There is also a cause effect issue as the lotion was not evenly applied and the color of the lotion is inconsistent. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: The woman did not squeeze the lotion into her hand.The movement of the woman while applying the lotion is not natural as her hand was deforming. There is also a cause effect issue as the lotion was not evenly applied and the color of the lotion is inconsistent.  I believe the video is AI-generated because of this discrepancy. .\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/67.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young adult sits at a desk in a well-lit room, facing a computer screen displaying a course on machine learning. The subject uses a mouse to navigate through the online course interface, scrolling and clicking on different sections. The background includes shelves with books and a lamp. The camera remains steady, focusing on the subject’s hands and the computer screen..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage symbols, not English. [Motion & Physical Consistency]: The deformed hand is unnatural. [Prompt Alignment]: The subject's face in the video is not visible, so we cannot identify whether it is a young adult. It is not a computer. The hand did not uses a mouse to navigate through the online course\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: No. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/68.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A woman with olive-toned skin and dark wavy hair is adjusting the buckles on her snowboard bindings, sitting on the icy ground at the peak of a mountain, while clouds drift around her against the backdrop of endless snowy peaks and blue sky..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Interaction with the object  as her thumb goes through is not in accordance to physical law. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/69.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman plays a beautiful violin solo, her bow moving gracefully across the strings as the music fills the room..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: NULL [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/70.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A fluffy golden retriever puppy, about two months old, runs energetically across a sunlit grassy plain. The grass is lush and gently sways in the breeze. Yellow, white, and pink wildflowers dot the landscape, and the sky is clear with a few clouds. The puppy chases a butterfly, jumping and running with its ears flapping and tail wagging. The camera captures the lively scene, highlighting the puppy’s playful movements and the bright, peaceful outdoor setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: There is another golden retriever puppy slightly visible which is not mentioned in the prompt.\nThe dog did not chases a butterfly.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/71.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The woman has a soft oval-shaped face with smooth, warm-toned skin. Her eyebrows are naturally full and slightly arched, framing almond-shaped brown eyes with long, dark lashes. She has a straight nose with a small hoop nose ring on the left side. Her lips are medium-full with a defined cupid’s bow, and her jawline is soft with a gentle taper toward the chin. Her hair is a sleek, blunt-cut bob in platinum blonde, parted slightly off-center, framing her face symmetrically. The overall impression is balanced and expressive, with a youthful, well-defined facial structure..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Good. [Prompt Alignment]: There is another ring in the middle of her nose.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/72.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: Under the warm amber hue of a summer sunset, a young man with curly red hair fries chicken wings over an open portable burner at a campsite, the golden oil bubbling energetically against the backdrop of a tranquil lake..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: The multiplying of the chicken wings, and the cause-and-effect of the disappearance of one chicken wings all seem unnatural. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/73.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A South Asian man sits at a kitchen table decorated with autumn leaves. He holds a pumpkin steady with one hand and uses a small carving knife in the other to carve intricate patterns into the pumpkin’s surface. The background shows warm lighting and seasonal decorations. The camera remains steady, focusing on his hands and the pumpkin as he works..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Yes [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: No - It looks like it was generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/74.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A simple office building transforms from neutral colors to bold red and white. The exterior gradually shifts walls turn red, accents and details become white, creating a sleek, branded look. \n\nProminently featured is the TurboTax logo: a red circle with a white checkmark. No text overlay. Emphasize the transformation and make the logo visible and accurate..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/75.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The subject, a Hispanic man, peels potatoes for dinner, the peels falling away as he prepares the vegetables for cooking..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: NULL [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/76.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A teenage South Asian boy leans against a wooden study table illuminated by a desk lamp. The camera zooms in on his elbow resting on the surface, highlighting the dry, textured skin as he lightly presses on it with his hand. The soft yellow light enhances the grooves and details of his skin. The text overlay 'if you have this' floats subtly over the frame. The background fades into darkness aside from the lamp's glow, pulling all attention to the elbow and skin's condition..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage symbols. [Motion & Physical Consistency]: There is distortion on the hand. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: No. Subject-Object Interaction: There is another hand visible from another subject. The text does not match. He did not press his elbow. There is distortion on the hand..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/77.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly East Asian man stands on a grassy field at sunrise, holding an American football in both hands. He demonstrates the throwing motion to a young boy standing beside him, showing how to grip and release the ball. The dewy grass glistens in the gentle morning light. The camera remains steady, capturing both subjects and the bright field in the background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: No [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: No - It looks like it was generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/78.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Middle Eastern man responds to a sidewalk interview in downtown, gripping the microphone and gesturing animatedly while his group listens and cheers him on..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The wordings are garbage symbols. [Motion & Physical Consistency]: There is stiffness in the hands, making it unnatural. There is deformity on the hand, making it unnatural. [Prompt Alignment]: He didn't grip the microphone.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/79.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Caucasian man stands at a stove in a bright kitchen. He holds a wooden spoon in his right hand and stirs vegetables in a frying pan, watching them cook in hot oil as they begin to soften. The background shows kitchen counters and utensils. The camera remains steady, focusing on the man and the pan..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: No [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: No - It looks like it was generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/80.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged man with dark curly hair and glasses, sitting on a wooden bench in a serene park at sunset, carefully packing tobacco into a smooth, mahogany smoking pipe, with golden light glinting off the pipe's polished surface..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: The cause-and-effect fails when smoke suddenly appears without a clear cause. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/81.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman with braided hair sits at a kitchen table. She holds a pair of scissors in her right hand and uses them to cut through the tape sealing a cardboard box. As she finishes cutting, the top flaps of the box open to reveal the contents inside. The background shows a bright kitchen with countertops and small appliances. The camera remains steady, focusing on her hands and the box..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Interaction with the object is not in accordance to physical law. Cause-effect failure as the mark of the scissor is not accurate. [Prompt Alignment]: Good.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/82.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Caucasian woman stands in a brightly lit room, holding a maraca in each hand. She shakes the maracas with energetic, rhythmic movements, matching the lively music playing in the background. The camera remains steady, capturing her upper body and the colorful room decor..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Yes [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: Yes - It looks real and not generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/83.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: POV from inside a shopping bag: A Black woman’s hand with glossy nude-painted nails reaches into the bag and grasps a boxed skincare product. The background shows the soft interior of the shopping bag, with diffused light filtering in. The camera remains steady, focusing on the hand and the product as it is picked up..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The wordings are garbage symbols. [Motion & Physical Consistency]: Good. [Prompt Alignment]: Good.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/84.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A whimsical underwater scene shows a mermaid with glowing blue-green scales standing on a sandy ocean floor. She holds a translucent, jellyfish-like bowling ball in both hands and rolls it toward a cluster of coral pins. Small bubbles rise around the ball as it moves forward. The background features colorful coral and gentle ocean currents. The camera remains steady, capturing the mermaid’s action and the underwater setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: The subject does not  rolls it toward a cluster of coral pins\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/85.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A fit middle-aged man stands in a gym locker room, wearing a sleeveless athletic shirt and shorts. He holds a tube of muscle gel in his right hand and squeezes it, causing a thick, clear gel to ooze onto his open left palm. The background shows lockers and benches under bright overhead lights. The camera remains steady, focusing on the man’s hands and the gel as it is dispensed..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The cause and effect of the gel coming out, even though he is not squeezing, is unnatural. [Prompt Alignment]: He is not squeezing.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/86.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A South Asian woman stands at a café counter holding a cup of coffee in her left hand. She pours steamed milk from a small metal pitcher into the coffee, carefully moving her hand to create a tulip-shaped pattern on the surface. The background shows a coffee machine and cups on the counter. The camera remains steady, focusing on her hands and the latte art as she finishes pouring..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: No [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: No - It looks like it was generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/87.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Hispanic woman stands at a kitchen counter. She holds a plate of ravioli in both hands and transfers the ravioli into a bowl. A slice of garlic bread rests beside the bowl on the counter. The background shows a bright kitchen with simple decor. The camera remains steady, focusing on her actions and the food preparation..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Good. [Prompt Alignment]: She didn't hold a plate of ravioli in both hands.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/88.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A close-up of a melting scoop of pistachio ice cream topped with crushed nuts, the creamy texture glistening under soft daylight..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: The cause and effect of the ice cream are unnatural because it does not melt naturally. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/89.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The subject, a Hispanic man, cuts a watermelon with a large knife, the sweet and refreshing fruit ready to serve..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: NULL [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/90.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young East Asian woman riding a motorized suitcase inside a shopping mall, passing clothing displays and retail booths..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage text. [Motion & Physical Consistency]: Good [Prompt Alignment]: It doesn't look like a real motorized suitcase.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/91.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Caucasian man stands in a well-lit room. He holds a perfume bottle in his right hand and sprays perfume onto his left wrist. Next, he uses his right hand to dab the perfume behind his left ear. The camera remains steady, capturing his actions and the neutral background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: His overall movements aren't natural. The effects of perfume spraying are unrealistic. [Prompt Alignment]: He held it in his left hand instead of right and he didn't spray perfume behind his ear.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/92.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A teenager vacuums a thick beige carpet, the vacuum lines creating neat, clean paths as sunlight pours in through a nearby window..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage Symbols. [Motion & Physical Consistency]: His left hand is distorted. [Prompt Alignment]: Good.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/93.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Caucasian man lies on the bed, relaxing after a workout, his eyes closed as he listens to music..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: NULL [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/94.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: In a fast-paced game scene, a middle-aged man with a stout figure and light skin, wearing a weathered blue jersey, releases an American football mid-air, the motion frozen for a second against a backdrop of roaring fans and colorful banners..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: The ball was spinning normally, then suddenly spun in the opposite direction. Stabilize to make it look normal. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/95.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged woman with auburn hair and freckled skin crouches beside a flickering campfire on a grassy meadow at sunset, carefully seasoning a pan of sizzling fish, the vibrant greenery blending seamlessly with the deep orange horizon..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: NULL [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/96.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Black man plays a bluesy tune on the harmonica, his eyes closed as he gets lost in the music..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: NULL [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/97.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged Slavic man wearing a blue jacket sits upright in the driver’s seat of a parked car. He faces forward, holding the steering wheel with both hands. The camera shows a medium shot of the car’s interior, with the dashboard and windshield visible. The camera slowly pans from left to right, capturing the man’s relaxed posture and subtle movements as he glances at the road ahead and adjusts his grip on the steering wheel..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: The subject is elderly\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/98.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Middle Eastern woman holds a sweater near her face and gently sniffs it, visibly pleased by the fresh, just-washed scent..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: There is a flickering issue with the sweater. [Motion & Physical Consistency]: The cause and effect fails because of the sleeve of the sweater. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/99.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A muscular Middle Eastern man with a dark beard stands on a metal ship deck beneath an overcast sky. He grips a lightweight steel throwing knife with a red stripe along the edge in his right hand, then swiftly throws it toward a moving robotic target. The blade strikes the target with precision. The camera remains steady, capturing the action and the industrial ship setting in the background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the subject movement as it was having stiffness and not smooth upon moving. Also with the subject interaction as the hand and the object was seen having a deformation. Lastly with the cause-effect as a knife or sword was seen flying on its own and it sudden disappears and in a while it sudden appears again making it appears unnatural. [Prompt Alignment]: The subject is not  swiftly throws it toward a moving robotic target.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/100.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman lifts a colorful shopping bag toward the camera with a wide grin, revealing neatly packed beauty and skincare products inside..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Good [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/101.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly South Asian man with a wispy gray beard, dressed in crisp linen pajamas, refreshes his bedroom at dawn. The room is bathed in soft, golden light filtering through grand arched windows. He spreads a luxurious silk comforter adorned with intricate patterns across a wooden canopy bed. His hands glide across the surface, smoothing wrinkles with intent precision. A gentle breeze rustles a nearby vase of fresh jasmine, adding an aromatic layer to the serene, elegant atmosphere of the classic bedroom setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: There is garbage text visible in the video. [Motion & Physical Consistency]: The movement of the man is not natural, as his hands move too stiffly. There is also a subject-object interaction issue, as the comforter becomes deformed while he is spreading it. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: The background does not appear dawn. The movement of the man is not natural, as his hands move too stiffly. There is also a subject-object interaction issue, as the comforter becomes deformed while he is spreading it. I believe the video is AI-generated because of this discrepancy. .\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/102.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A man and a woman are on a sunny rooftop. The man stands beside a row of black and silver solar panels, holding a spray hose and spraying water onto the panel in front of him. The wind blows the water across the rooftop, soaking the woman standing nearby. She stands still, looking at the man with a raised eyebrow. The man continues spraying water, while the woman watches him. The camera remains steady, showing both subjects and the rooftop in daylight..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Their faces are deformed and the effects of the water aren't consistent and realistic. [Prompt Alignment]: It wasn't shown that the woman was soaking in water.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/103.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman stands in her backyard, holding the handle of a lawn mower. She steadily pushes the mower forward, cutting through the grass as the machine buzzes. The background shows a wooden fence, green shrubs, and sunlight filtering through nearby trees. The camera remains steady, capturing her movement and the freshly cut lawn..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Yes [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No - The video does contain nudity or harmful content. Cause-Effect: No - It looks like it was generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/104.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A mature White man with a slim build sits in the driver’s seat of a car, wearing a red polo shirt. The car’s interior, including seats and a window, is visible in the background. He looks steadily at the camera and speaks in a calm, soft voice, his posture relaxed. The camera remains steady, capturing his composed expression and peaceful demeanor as he talks..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: He doesn't look slim.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/105.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Caucasian woman stands in a well-lit room. She raises one arm and bends it behind her head, then leans her upper body to the side to deepen the stretch. The camera remains steady, showing her full posture and the surrounding space..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Yes [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: Yes - It looks real and not generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/106.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged East Asian woman with short, sleek hair is seated behind a modern news desk in a brightly lit studio, pointing at a stack of notes while adjusting her microphone pinned to her blazer, framed by glowing screens displaying breaking news headlines..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The overlay wordings are just garbage, not English. [Motion & Physical Consistency]: The hand movement is not natural. [Prompt Alignment]: She did not adjust her microphone.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/107.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Latina woman with a high ponytail and athletic build stands in her bedroom, the walls decorated with contemporary art. She wears a sleeveless purple crop top paired with denim shorts. Holding a sleek black microphone, she gestures enthusiastically toward a standing rack filled with colorful clothing. The room has a relaxed, modern vibe, with soft lighting casting shadows on the clothes, creating a comfortable yet fashionable setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Her overall movements are unnatural. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: She isn't a Latina, she is more likely a black woman..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/108.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Hispanic woman sits on the sand at a beach in the early evening. She holds a long skewer with a marshmallow over a small campfire, slowly rotating it to roast the marshmallow. In the background, ocean waves crash gently along the shore. The camera remains steady, capturing the woman’s actions and the beach setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Yes [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: Yes - It looks real and not generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/109.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly South Asian man works in a dimly lit garage, gripping a sturdy, red power drill to secure metal bolts on an old bicycle frame. Shelves of tools line the walls, and background noise of distant crickets fills the air..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: There is garbage text visible in the drill. [Motion & Physical Consistency]: The movement of the man appears stiff. There is an object-object interaction issue as the drill passes through the bolts. There is also a cause effect issue as the bolt suddenly appeared in the metal and the object behind him is moving on its own. [Prompt Alignment]: The man is not South Asian. The man does not appear secure metal bolts.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/110.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Latino man with a muscular build stands confidently in front of a makeup table indoors. He wears a dark t-shirt and maintains a relaxed posture while speaking directly to the camera, keeping steady eye contact and a natural, satisfied expression. The background features a woman in a white top with her face blurred, appearing to apply something to her face. The atmosphere is neutral and suggests a personal care setting. The camera remains fixed at eye level, capturing the man’s composed and steady delivery..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Can't determine if he's standing in front of a makeup table.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/111.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged man with short dark hair sits at a desk in a dimly lit room. He works at a computer with two monitors displaying code. The desk holds a white mug, a notebook, a pen, and a computer mouse. A window behind him shows the night sky and faint outdoor lights. The camera remains steady, capturing the man’s focused work and the surrounding workspace..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Cause-effect failure as the screen has no changes with all the typing. [Prompt Alignment]: No mouse.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/112.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Latino woman with long, sleek brown hair stands in a modern kitchen. She wears a modest short-sleeve blouse and jeans. The background features a refrigerator and kitchen items, creating a casual, relaxed atmosphere. She faces the camera, maintaining steady eye contact and a composed posture. Her facial expression shows subtle dissatisfaction, with slight tension around her mouth and a serious look in her eyes. She speaks calmly and firmly, expressing discontent through her tone and natural facial cues. The camera remains fixed, capturing her upper body and the kitchen setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: textural issues in lips [Motion & Physical Consistency]: good [Prompt Alignment]: woman looks indian rather than latino\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/113.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Eastern European man with medium-long blonde hair stands in front of a villa, wearing a dark polo shirt. He faces the camera with his upper body exposed, speaking directly to the lens. As he talks, he occasionally gestures lightly with his hands or shifts his posture. The background shows the villa’s exterior and surrounding greenery. The camera remains steady, focusing on the man’s expressive movements and the elegant setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The texture of the hand is flickering. [Motion & Physical Consistency]: The hand deformed. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/114.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged Arabic woman stands in a quiet library, shown from the waist up in a medium shot. She faces the camera directly, maintaining steady eye contact as she speaks clearly and confidently. She wears a tailored, earth-toned blouse with a mid-coverage neckline, covered shoulders, and subtle stitching, paired with minimal accessories for a professional look. The background features softly blurred tall bookshelves, a long wooden reading table with chairs, a desk lamp, a globe on a corner table, and large windows letting in natural light. The camera remains steady at her eye level. She stands calmly, addressing the viewer without gestures..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: The subject is not middle-aged\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/115.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A teenage girl with circular glasses and a yellow hoodie sits at a desk in a dark room, illuminated only by the glow of her computer monitor and tower. She moves a black mouse smoothly on a patterned mat and types quickly on a colorful keyboard while playing a vibrant, busy game on the screen. The camera remains steady, capturing her focused expression and the soft light from the computer..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Hand movement is not smooth and lacks fluidity. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/116.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: a man is having trouble cleaning his kitchen counter, he looks helpless.\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Movement is not smooth and lacks fluidity. Deformation is visible in his left hand and also a yellow liquid suddenly appears on the table. [Prompt Alignment]: There is a white object he uses in cleaning but was not mentioned in the prompt.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/117.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly Hispanic man sits in a cozy bedroom with warm, soft lighting. He gently reaches out and lightly touches a small box placed in front of him. Wooden furniture and a neatly made bed are visible in the background. The camera remains steady, capturing his subtle movement and the inviting atmosphere of the room..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/118.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Caucasian man in a crisp black suit walks along the Great Wall of China, the camera capturing the full-body shot as he strolls, showcasing the elegance of his outfit against the awe-inspiring ancient structure..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The video is blurry. [Motion & Physical Consistency]: Good. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: I think it's generated by AI\n.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/119.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Caucasian woman uses a stylus to navigate her tablet screen, drawing shapes and creating digital art with precise strokes..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The mark is not accurate and inconsistent. [Prompt Alignment]: The face is not visible.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/120.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young White man wearing a short-sleeve athletic shirt, shorts, and soccer cleats stands on an outdoor sports field in front of a goal. He dribbles a soccer ball forward with his right foot, then quickly strikes the ball toward the goal using the inside of his foot. His arms move naturally for balance as he completes the shot. The background shows green grass, the goal net, and distant trees under bright daylight. The camera remains steady, capturing the entire drill from a side angle..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: The overall movement, the deformation, and the cause-and-effect of the ball all seem unnatural. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/121.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Middle Eastern woman gently applies lotion to her elbows and forearms, blending it in slowly as she relaxes in a softly lit bedroom..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The movement appears stiff and there is a slight flick on the finger. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: The movement appears stiff and there is a slight flick on the finger..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/122.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Southeast Asian man scoops steaming rice and curry with a spoon, the aroma rising as he eats in a warmly lit dining room..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The cause-effect of the rice is unnatural. [Prompt Alignment]: He did not  scoops steaming rice and curry with a spoon.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/123.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Malay-Japanese man with long hair and an athletic build stands in a brightly lit clinic. He holds a probiotic bottle in one hand, his face displaying a startled expression as he looks directly at the camera. The background features clean white walls and clinical equipment. The camera stays steady in vertical 9:16 format, capturing the vivid, animated style and fine details of the scene..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Texture issues on the hair and on the bottle. [Motion & Physical Consistency]: His teeth are distorted as he speaks making it unnatural. [Prompt Alignment]: Good.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/124.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A South Asian woman with medium-length hair rinses the shampoo from her hair. The camera captures the foamy bubbles dissolving and flowing down her hair, revealing freshly washed, shiny hair as the water rinses it clean..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Cause-effect failure as the hair is wet as the water flows. [Prompt Alignment]: The face is not visible.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/125.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young woman with straight, dark hair in a ponytail wears athletic leggings and a fitted tank top. She runs on a treadmill in a well-lit home gym. A pink shopping bag dangles in midair several feet in front of her, suspended from the ceiling by a string. She glances at the bag occasionally as she runs at a steady pace. The camera remains fixed, capturing her movement and the gym setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage symbols, not English. [Motion & Physical Consistency]: The deformed hand and the cause and effect of the text in the treadmill doesn't match with the reflection [Prompt Alignment]: Cannot determined if she  glances at the bag occasionally\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/126.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged East Asian woman in a flowing emerald green dress steps out of an enchanted forest bookstore. The air shimmers with soft gold light, and enchanted books hover on translucent shelves, fading into the background as she exits. Her elegant updo and intricately designed leather shopping bags, embossed with glowing runes, catch the last golden rays of sunlight, adding a whimsical charm to the magical setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The subject's hand and the unrealistic, sudden disappearance of the bookshelves are unnatural. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: I think it's generated by AI because of the unnatural movement..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/127.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A woman with long, pale pink painted fingernails walks along a sunlit path lined with colorful wildflowers. Sunlight filters through soft, hazy clouds, casting gentle shadows. Butterflies flit among the flowers as she moves forward. The background features a glowing, celestial sky with a faint rainbow arc. The camera follows her at a steady pace, capturing the whimsical, enchanted atmosphere..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The passing through of the butterflies and the sudden appearance of them in the side seem unnatural. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/128.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman with curly hair wears a pair of sleek, reflective sunglasses as she leans confidently against a weathered brick wall in a sunlit urban alley. The vibrant graffiti behind her adds a burst of color, while her cool demeanor radiates through her subtle yet commanding pose, the sunlight glinting off the lenses of her shades..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Good [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/129.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A late 20s White man with dark hair and a slightly chubby face stands at a campsite, dressed in an office suit. He holds a tennis ball in his right hand, quickly jumps once in place, then stands still and looks around with a bewildered expression. The background shows tents and camping gear scattered around on dry grass under daylight. The camera remains steady, capturing his movements and the campsite setting.man is not talking.\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Movement is not smooth and lacks fluidity. Deformation is visible in his right arm as it appears cut. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/130.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly East Asian woman with short silver hair sits at a vintage wooden desk in a dimly lit study. The warm glow of a desk lamp highlights her hands as she unwraps delicate floral wrapping paper, revealing a soft, neatly folded burgundy cashmere scarf. She gently runs her fingers over the scarf’s surface, her eyes reflecting quiet joy. The background shows shelves of old books and dark wood paneling. The camera remains steady, focusing on her actions and the cozy atmosphere..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Interaction with the wrapper as it deforms is not in accordance to physical law. Cause-effect failure as some part of the wrapper moves on its own. [Prompt Alignment]: She didn't unwrap the floral wrapping paper.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/131.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged Latina woman in a flowing scarf confidently dons a pair of sleek, cat-eye sunglasses while standing on the golden sands of a quiet desert at dusk. The orange-pink hues of the setting sun cast a warm glow over her as she elegantly places her hands on her hips, striking a bold yet serene silhouette against the expansive landscape..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The movement of the woman is not natural as her finger on her right hand deformed. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: The movement of the woman is not natural as her finger on her right hand deformed. I believe the video is AI-generated because of this discrepancy..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/132.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young woman with long brown hair, wearing athletic leggings and a fitted tank top, stands in a well-lit gym with weight racks and mirrors in the background. Holding a dumbbell in each hand at her sides, she bends at the hips and knees, lowering her torso while keeping her back straight. She then straightens her legs and lifts her torso, raising the dumbbells to waist height. She pauses briefly in the upright position, then lowers her torso again by bending at the hips and knees, repeating the deadlift motion. The camera remains steady, capturing her full body and the gym environment..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Good [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/133.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The subject, a Hispanic woman, stirs her sparkling water with a straw, enjoying the fizz as it mingles with the lemon zest..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The effects of lemon zest when stirred aren't consistent and realistic. [Prompt Alignment]: Her face isn't visible to determine if she is Hispanic.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/134.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Black man in a casual shirt and jeans salsa dances with a woman in a colorful dress on a polished dance floor. He spins her around his body, lifting her leg and guiding her movements with his arm. His footwork is quick and coordinated as he turns and steps, while she wraps her arms around his neck and shifts her weight in response. The background shows a lively indoor venue with string lights and other dancers. The camera remains steady, capturing their synchronized movements and the energetic atmosphere..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is a subject-subject interaction issue as the legs of the man was deforming while dancing becoming excessive as well as the people in the background. [Prompt Alignment]: The man did not lift her leg and the woman did not wraps her arms around his neck.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/135.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Latina woman with long, dark wavy hair and medium skin tone sits in the driver’s seat of a car. She wears a grey buttoned top and a silver cross necklace. She looks directly into the camera, maintaining steady eye contact as she speaks with a subtle, curious expression. The car’s seat and headrest are visible behind her, with daylight and outdoor scenery seen through the window. The camera remains steady, capturing her upper body and face in a relaxed, casual setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/136.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A white teenage girl with straight blonde hair sits cross-legged on her bed in a softly lit bedroom. She wears a light blue T-shirt and shorts. Holding a bottle of creamy body lotion in her right hand, she dispenses some onto her left palm, then rubs the lotion onto her legs using both hands in smooth, circular motions. The camera remains steady, showing her focused on applying the lotion as it gradually absorbs into her fair skin. The background includes pillows and a nightstand..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Movement is not smooth and lacks fluidity. [Prompt Alignment]: She holds a bottle of lotion in her left hand.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/137.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A woman with painted nails lifts a veggie-loaded pizza slice, and strings of melted cheese stretch from the slice to the pan below..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: There is garbage text visible in the shirt. [Motion & Physical Consistency]: There is a cause effect issue as the stretched cheese suddenly disappeared. [Prompt Alignment]: There is no pan visible in the video.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/138.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A tattooed woman stands in a dark, abandoned cathedral filled with cold smoke and cinder mist. She wears black leather and metal accessories, her figure partially shrouded in shadow. Her arms are raised toward the sky in a ritual posture, invoking a gothic ceremony amid scattered black candles, broken crucifixes, and hanging chains. The setting is lit only by deep black and dark silver tones—no warm light present. Dust hangs in the air as the camera moves slowly and hypnotically, capturing high-contrast cinematic grain reminiscent of 90s film. The mood is grim and silent, evoking a dark, nihilistic post-apocalypse with an atmosphere of black romanticism..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/139.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black man in a navy jacket ties a strong knot in a fishing line while standing on a narrow boat at sunrise, his reflection appearing in the glistening, calm water beneath, surrounded by faint mist from the early morning air..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage [Motion & Physical Consistency]: The movement is not smooth and looks stiff. The fishing line seems deformed and passes through the man's hand. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/140.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The video shows the interior of a moving car from the perspective of the back seat, focusing on the driver’s seat, dashboard, and the road ahead. A young Brown man sits in the driver’s seat with both hands resting on his thighs, not touching the steering wheel. The car moves forward on a two-lane road and passes under an overpass. The dashboard displays illuminated controls, and a mounted device above the dashboard shows a navigation map. The car’s built-in screen also displays a navigation interface. The man keeps his hands off the steering wheel as the vehicle continues to drive straight. The camera remains steady throughout the scene..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: The cause and effect of the reflection from the mirror is unnatural. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/141.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: 一个阿拉伯男性，头戴白色的头巾，年龄在30-40岁左右，面对镜头说话，表情轻松自信；上半身出现在镜头中，背后是阿拉伯人的客厅.\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/142.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman shakes the maracas in rhythm, the sound of the shaking creating a lively beat as she dances..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: NULL [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/143.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Middle Eastern woman holds a glass serum dropper over her cheek, releasing a droplet that lands precisely on a patch of uneven skin texture..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Cause-effect failure as the product dispenses inaccurately. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/144.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Hispanic man stands in a small boat on a lake. He grips a fishing rod with both hands and pulls a large, splashing fish up from the water and into the boat. The background shows calm water and distant trees. The camera remains steady, capturing the man’s actions and the surrounding scenery..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Yes [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: Yes - It looks real and not generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/145.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An aerial view shows a busy food festival during the daytime. A large crowd of adults walks around the area, moving between food stalls and tents. Several food trucks are parked nearby, each displaying different types of food. The background includes open space, event banners, and bright daylight. The camera remains steady, capturing the lively atmosphere and movement of people and food stalls..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The video has a garbage caption. [Motion & Physical Consistency]: There is an issue with the subject movement as the motion of the subject was not smooth and lacks fluidity. Also with the subject interaction as the body of the subject was seen having a deformation. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/146.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Caucasian woman inserts her credit card into a machine, entering her pin code before finalizing her payment..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Good [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/147.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged man with light brown skin and salt-and-pepper hair, wearing a crisp white shirt, steps back on the sandy beach at sunset where waves lap at the shore behind him. He arches his torso slightly, creating a natural curve that contrasts with the peaceful horizon. His relaxed arms and open stance project calm confidence. He fixes the camera with a kind expression, then gently blows a kiss, the wind carrying a sense of warmth as the golden light envelops the scene..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The body movement is not natural when he bends in. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: He did not blow a kiss..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/148.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An overhead shot shows an adult’s hands holding a sunscreen tube above their forearm, which rests on a colorful beach towel. The person presses the tube, dispensing a smooth stream of lotion onto their forearm. The background features the textured pattern of the beach towel. The camera remains steady, clearly capturing the action from above..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage [Motion & Physical Consistency]: There is an extra arm. The sunscreen being applied or dispensed is not consistent and not accurate. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/149.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Caucasian woman performs a backflip on a trampoline, her body soaring through the air before she lands softly on the mat..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Subject movement is unnatural. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Unrealistic movements..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/150.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Black woman with curly hair applies moisturizing cream to her face in a bright, modern bathroom. The camera focuses on her hands as she gently massages the cream into her skin, revealing a smooth, hydrated complexion..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is a cause-effect failure. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: There is a cause-effect failure..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/151.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Southeast Asian man carefully lifts a thick, saucy piece of lasagna onto a plate, the layers of pasta, meat, and cheese clearly visible in the close-up..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the cause-effect as the lasagna was dripping inconsistently and as the large portion was drip it goes back to the same position and portion of the lasagna [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: There is an issue with the cause-effect as the lasagna was dripping inconsistently and as the large portion was drip it goes back to the same position and portion of the lasagna.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/152.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A South Asian man sits on a bench in a well-lit bedroom. He picks up a pair of black leather dress shoes and slides his feet into them, adjusting each shoe for a comfortable fit. The camera remains steady, capturing his actions and the tidy bedroom background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: No [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: No - It looks like it was generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/153.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young man with dark brown skin and short curly hair stands inside a futuristic holographic arena. He performs a smooth cheerleading routine, twirling neon pom-poms that shimmer and flicker in his hands. The digital crowd in the background glows with animated lights and cheers. The camera remains steady, capturing his coordinated movements and the vibrant arena setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: The subject's body is deformed and the movement is unnatural. Stabilize to make it look natural. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/154.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman holds a freshly washed shirt to her face and smiles softly as she inhales its clean scent..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Good. [Prompt Alignment]: Good.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/155.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged woman with a scarf and glasses opens a water bottle while sitting at a cafe patio. The camera zooms in on her hand as she twists the cap, revealing the refreshing water inside as she enjoys a moment of relaxation in the fresh air..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage text. [Motion & Physical Consistency]: Her overall movements are unnatural. Interaction with the water bottle is unnatural as well. The cap disappears unnaturally. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: She didn't twist the cap open the cap unrealistically disappears..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/156.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: In a whimsical enchanted forest with glowing mushrooms and fireflies, a little girl with dark brown skin zips between giant trees on roller skates designed to look like they’re carved from wood. Her laughter lights up the scene as the skates leave sparkling trails behind her..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: There is a flickering issue with the trees. [Motion & Physical Consistency]: The overall movement,  the interaction with the roller skates and the cause and effect of the water are all unnatural. [Prompt Alignment]: There were no sparkling trails\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/157.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young South Asian woman in a lush green park sits cross-legged with a digital tablet on her lap, using a stylus to paint a digital landscape of a mountain, her face glowing from the device's screen in the evening light..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The subject's movement and the interaction between the subject and object is unnatural. There is a cause-effect failure. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: The subject's movement and the interaction between the subject and object is unnatural. There is a cause-effect failure..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/158.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The subject, a South Asian woman, enjoys a bite of strawberry shortcake, the fresh berries adding a refreshing sweetness to the dessert..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: NULL [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/159.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: In a dream-like underwater setting, an elderly South Asian man floats serenely, his hands gliding over his dry and wrinkled elbow. His tan skin is illuminated by shimmering blue light filtering through the water above. The camera zooms in as text overlay forms along the curves of his visible hair follicles, quietly asking, 'if you have this,' blending with the surreal aquatic atmosphere..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The wordings are garbage symbols. [Motion & Physical Consistency]: Good. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: No. Subject-Object Interaction: The subject is not South Asian. The text does not match. He didn't glide his hand into his elbow..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/160.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Southeast Asian woman with long black hair and a bright smile waves both hands enthusiastically toward the camera while sitting in a softly lit bedroom..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The video has a  garbage caption. [Motion & Physical Consistency]: There is an issue with the subject movement as the subject was waving too fast and it has distortion as the subject waves her hand. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: There is an issue with the subject movement as the subject was waving too fast and it has distortion as the subject waves her hand..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/161.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman stands in a serene garden at sunset, examining her arm. The camera zooms in as her fingertips lightly glide over a patch of dry, textured skin on her elbow. Soft, golden light enhances the contrast between smooth and rough areas. A text overlay appears on the screen: 'Is this your skin?' The focus stays on the intricate detail of the skin and the gentle touch of her hand, set against the blurred greenery of the background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage text. [Motion & Physical Consistency]: Subject hand movement is unnatural. Cause-effect fails. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: The overlay text is inaccurate. She is not examining her elbow..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/162.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The subject, a Hispanic woman, carefully opens a package with a cutter, revealing the contents inside to examine them..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: There is garbage text visible in the video. [Motion & Physical Consistency]: The movement of the woman appears stiff. There is an object interaction as the wrapper was deformed. There is also a cause effect issue as the package was opened without proper motion. [Prompt Alignment]: There is no cutter visible. The woman did not open the package with a cutter.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/163.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: In a quiet forest, a South Asian man with a beard opens a can of lime soda. The camera focuses on his hand as he pulls the tab, capturing the fizzy bubbles rising to the top, with the fresh, clear soda inside..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage symbols, not English. [Motion & Physical Consistency]: The hand movement, the deformation, and the cause-and-effect of the soda all seem unnatural. [Prompt Alignment]: There is another hand  in the video which is not mentioned in the prompt and he did not pulls the tab.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/164.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Caucasian woman with shoulder-length hair stands in a well-lit bedroom. She gathers her hair and twists it up into a bun, then secures it neatly with a hair clip. She uses her hands to adjust a few loose strands around the bun. The camera remains steady, capturing her actions and the tidy bedroom background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The hand and the hair and clip get deformed. [Prompt Alignment]: She did not twist it up in the video.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/165.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: In a tranquil forest clearing bathed in golden morning light, a young Black woman with coiled hair gently examines her arm, revealing a patch of visibly dry, textured skin near her elbow. Her fingers lightly graze the area, showcasing a tender interaction. The neutral background features soft green and brown hues, while text overlay reads 'notice this?' in subtle, translucent lettering. The focus remains on the rich contrast between her skin and the serene setting, emphasizing the details of the texture and motion..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: There is garbage text visible in the video. [Motion & Physical Consistency]: Good [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: The overlay text is incorrect; it is not 'notice this?'. The woman does not examine her elbow, and the textured, dry skin is missing from the elbow area. Additionally, garbage text is visible in the video. These discrepancies suggest that the video is AI-generated..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/166.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman with natural curly hair sits at a vanity in a softly lit bedroom. She holds a makeup brush in her right hand and applies bronzer to her cheekbones, blending the product in gentle, circular motions. The camera remains steady, capturing her focused expression and the vanity background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Yes [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No - The video does contain nudity or harmful content. Cause-Effect: No - It looks like it was generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/167.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Arabic woman wearing a white t-shirt and a long red daisy-patterned skirt walks through an airport terminal, holding a backpack. She holds a smartphone in her right hand, recording herself from her perspective. An Arabic man in a short-sleeved cream shirt and jeans walks beside her, pulling a suitcase on wheels. Both are smiling and looking toward the camera. The background shows airport seating and overhead bins under bright lighting. The camera remains steady, capturing their coordinated movement through the terminal..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: good [Motion & Physical Consistency]: subject missing pinky finger [Prompt Alignment]: subject holding smartphone on her left hand.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/168.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Samsung laptop sits on a light wood table in a well-lit room with yellow and white walls. The laptop’s screen displays a dynamic, abstract design with blue and orange tones. The camera slowly pans across the laptop, capturing its front and back as it rotates 360 degrees. The background includes a white wall with a vertical strip of yellow paint, a white doorframe, and a partially visible white box on the floor. The camera remains steady throughout the sequence..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage symbols [Motion & Physical Consistency]: Good [Prompt Alignment]: The laptop is not a Samsung brand. Did not rotate 360 degrees\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/169.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A South Asian man in a light-colored shirt stands at a kitchen table decorated for a festive breakfast. He sets a bowl of fruit salad on the table, then uses his hands to gently arrange the fruit pieces inside the bowl. The background shows bright morning light and festive tableware. The camera remains steady, capturing his careful movements and the cheerful setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: There is deformity in the finger. Cause-effect fails as some fruit in the salad appears and disappears and some seem to move on its own. [Prompt Alignment]: He is not South Asian.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/170.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly Black man with a graying beard sits on a wooden bench in a quiet park filled with autumn leaves. He carefully opens a matte black drone box, lifts the drone out, and examines its intricate details as sunlight reflects off its polished surface. The camera remains steady, capturing his focused expression and the peaceful park background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is a subject-object interaction issue as the box was deformed and passes through his hand. [Prompt Alignment]: He did not carefully opens a matte black drone box, lifts the drone out, and examines its intricate details as sunlight reflects off its polished surface.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/171.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: Orange and lemon slices rain down through soft natural light, some catching the sun as they tumble, their juicy texture highlighted in mid-flight..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the object object interaction as the orange was seen transform into lemon and again it back to the orange. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: There is an issue with the object object interaction as the orange was seen transform into lemon and again it back to the orange. .\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/172.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A serene indoor scene where a petite elderly woman with pale skin and a braided silver bun milks a cow by hand, softly humming to herself as sunlight streams through a nearby window, catching floating dust motes and the gleaming surface of the metal bucket..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: The prompt mentioned a cow but it was not visible in the video. The prompts also mentioned she milks a cow by hand but was not visible in the video.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/173.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly South Asian man with salt-and-pepper hair, dressed in a neat button-up shirt, sits at a wooden coffee shop table. He lifts his arm, carefully tilting his classic analog watch with a deep brown leather strap, squinting slightly to read the small black numbers under the soft glow of hanging lights..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/174.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Hispanic woman with long dark hair stands at a vanity table in a softly lit room. She holds a bronzer compact in one hand and uses her other hand to lift open the lid, revealing the finely pressed powder inside. She tilts the compact slightly so the light highlights the powder’s surface. The camera remains steady, focusing on her hands and the compact against the vanity background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: The lid moves up on its own. [Prompt Alignment]: She didn't use her other hand to lift open the lid.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/175.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A hand moves across a dark marbled countertop, with a white paper towel holder visible in the background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/176.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: Two Black teenagers stand side by side outdoors on a sunny afternoon, each holding a smoothie bottle. They smile and bump their bottles together in a playful cheers, then both take a long sip while looking at each other. The background shows green trees and bright natural light. The camera remains steady, capturing their cheerful interaction..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The movement is not smooth and lacks fluidity. The straw pierced through their mouths. The amount of smoothie of the subject in the right is inconsistent as it suddenly decreases before sipping. [Prompt Alignment]: They are not looking at each other while sipping.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/177.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A man with medium brown skin secures a weathered fishing boat to a wooden dock on a quiet lake at twilight. A young boy with medium brown skin stands nearby, watching the man intently. The background features distant cypress trees and a sky filled with warm evening light. The camera remains steady, capturing both subjects and the peaceful lakeside setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Deformed hand and unnatural movement. Stabilize to make it look normal. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/178.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A teenaged South Asian boy stands confidently in a serene forest clearing during twilight, surrounded by tall trees. He holds a green patterned playing card in his right hand and throws it sharply at a rope hanging nearby, slicing through the rope with precise accuracy. The camera remains steady, capturing the boy’s focused action and the tranquil forest setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: No [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: No - It looks like it was generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/179.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The subject, a Hispanic woman, shakes a bottle of sunscreen before applying it to her skin, ensuring a well-mixed formula..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The wordings are garbage symbols. [Motion & Physical Consistency]: There is stiffness in the hands. Cause-effect failure as the application of sunscreen is not accurate. [Prompt Alignment]: She's not  Hispanic. She didn't shake the bottle.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/180.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A close-up shot shows a Settlers of Catan board game set up on a wooden table. A light-skinned adult hand rolls two dice onto the table near the game board, then briefly touches the dice to adjust their position. The background includes visible game pieces and cards arranged around the board. The camera remains steady, focusing on the hand and dice during the action..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: The hand movement is not natural and the subject-object interaction fails because the object deform.\nThere is some changes of the dice. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/181.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Filipino grandmother stands in a brightly lit kitchen, holding a plate filled with large, juicy pieces of cooked chicken that are still steaming. She gently lifts the plate with both hands, displaying the chicken’s golden skin and tender meat. The background shows kitchen counters and utensils. The camera remains steady, focusing on the grandmother and the plate of chicken..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: The subject's face is not visible.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/182.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman with neatly manicured fingers holds the pastel pink spatula, spreading a thick, glossy white cream across a reflective glass surface. The camera captures her deliberate sweeping motions from above, focusing on the creamy material's texture and the spatula's smooth glides. The substance reflects soft daylight streaming through an unseen window, adding a warm glow to the scene. Minimalistic and clean, the video highlights the perfect uniformity of the cream as gentle ripples form with each pass of the spatula..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: She did not touches the other part of the cream but there is a traces of the spatula on it. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: The subject's face is not visible we cannot identify if she's young or not..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/183.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: Under a golden afternoon sky, an elderly East Asian man, wearing a charcoal beret and a long beige coat, stands on the sidewalk of a quaint seaside town. As the camera glides along the line of people, he extends his hand to pluck a pink-wrapped bouquet from a charming vehicle parked nearby. The ocean glimmers faintly in the distance as he examines the flowers, with the vibrant pink signage standing out against the cobblestone street. Seagulls flutter above, adding to the peaceful yet lively atmosphere..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage symbols. [Motion & Physical Consistency]: The way the seagulls fly is not natural. The subject-subject interaction fails because one seagull looks like it pass through another seagull. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: The way the seagulls fly is not natural. The subject-subject interaction fails because one seagull looks like it pass through another seagull..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/184.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Asian woman with visible large pores on her face stands facing the camera. She gently turns her head from side to side, allowing the camera to capture the texture and surface of her skin. The background is softly lit and neutral. The camera remains steady, focusing on her face and subtle head movements..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Texture lips [Motion & Physical Consistency]: Good [Prompt Alignment]: There's no visible large pores\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/185.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman with curly hair stands in front of a brightly lit mirror. She uses a small wooden spatula to carefully apply a thin layer of golden wax onto her eyebrow, maintaining a calm and focused expression. The camera remains steady, capturing her precise motion and the well-lit bathroom background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Texture hair. [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/186.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman with braided hair stands on a city sidewalk, holding a jump rope in both hands. She swings the rope overhead and begins jumping rope, moving her feet quickly as the rope passes under her legs. Her arms and legs alternate in a steady rhythm, causing her hair to swing with each motion. The background shows passing pedestrians and storefronts. The camera remains steady, capturing her energetic skipping and the dynamic movement of the rope..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Her overall movement is stiff and not fluid, making it unnatural. Interaction fails as there are deformities in the appearance of the people passing by. [Prompt Alignment]: The subject-object interaction in the prompt is inaccurate as she didn't move her feet quickly as the rope passes under her legs.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/187.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Middle Eastern man with a well-groomed beard vacuums the hardwood floor in his living room with a robotic vacuum cleaner. The camera zooms in on the vacuum as it effortlessly glides across the floor, picking up dust and debris from every corner of the room..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The wordings are garbage symbols. [Motion & Physical Consistency]: Interaction with the object is not in accordance to physical law. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: No. Object Existence: No. Subject-Object Interaction: The subject is not clearly visible. The vacuum is not a robotic vacuum. The vacuum didn't effortlessly glide across the floor, picking up dust and debris from every corner of the room..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/188.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: Amid the dynamic chaos of a high-tech training facility with holographic screens showing game stats, a muscular Black man in his late 30s wearing a blue-and-white jersey bounces a glossy white soccer ball once before sending it in a clean arc toward an invisible teammate..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: The cause-and-effect fails due to the ball bouncing unnaturally. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/189.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Hispanic woman stands in a brightly lit room. She raises both arms wide and cheers with excitement, her face showing enthusiasm. The camera remains steady, capturing her celebratory gesture and expression..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Yes [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: No - It looks like it was generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/190.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A white teenage girl with straight blonde hair sits cross-legged on her bed in a softly lit bedroom. She wears a light blue T-shirt and shorts. Holding a bottle of creamy body lotion, she dispenses some onto her hands and gently rubs the lotion onto her legs in smooth, circular motions. The camera remains steady, showing her focused application and the cozy bedroom setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Her overall movements aren't natural and realistic. Her hand deforms unnaturally and when he puts down the bottle there is still her fingers attached to it and the effects of the lotion are unnatural and unrealistic. [Prompt Alignment]: She isn't sitting cross-legged.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/191.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: On an isolated desert planet, a futuristic sailboat with translucent, glowing sails hovers over shifting golden sands, piloted by a woman with olive skin wearing a metallic silver suit, the sun blazing red behind her..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The subject's movements, interactions, and cause-and-effect are all unnatural. Make them look more natural. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: I think it's generated by AI because of the unnatural movement..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/192.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman in a purple swimsuit dives into a hotel pool, jumping from the side. The camera focuses on her form in mid-air, capturing her as she gracefully enters the water, her splash rippling across the surface of the pool..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: There is stiffness in her movements. [Prompt Alignment]: Good.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/193.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A college-aged woman with long dark hair stands beside a messy bed in a small dorm room. She holds a closed laptop in one hand and drops it onto the bed, where it lands among scattered notebooks and tangled sheets. The laptop lid bounces slightly upon impact. The background shows a cluttered nightstand and posters on the wall. The camera remains steady, capturing the action and the disorganized setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the subject interaction as the hand was having a deformation and it went through to the object. [Prompt Alignment]: The subject is not clearly visible. The laptop lid does not  bounces slightly upon impact.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/194.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The subject, a Hispanic woman, performs jumping jacks in the park, her arms and legs moving in unison as she enjoys the outdoor exercise..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: The movement of the subject appears not smooth and fluid. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/195.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Black woman with natural curls walks across a coffee shop in distressed light blue jeans. The camera focuses on her legs as she passes by, showcasing the comfort and style of the jeans. (40 words).\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The subject's leg suddenly interchange. [Prompt Alignment]: The subject's face is not visible.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/196.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Black man with short hair opens a red and white cookie packet in his cozy kitchen. The camera zooms in on his hands as he rips open the packet, revealing delicious chocolate chip cookies inside, ready to enjoy..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: There is garbage text visible in the packet. [Motion & Physical Consistency]: The man's movement and interaction while opening the packet appear unnatural, with stiff hands and the packet deforming during handling. Additionally, the packet opens unnaturally without proper action. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: The man did not open the packet properly as expected. The man's movement and interaction while opening the packet appear unnatural, with stiff hands and the packet deforming during handling. Additionally, the packet opens unnaturally without proper action. I believe the video is AI-generated because of this discrepancy..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/197.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly East Asian man presses a button to raise his desk and begins sorting papers in a quiet, book-lined room..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Movement appears stiff and not smooth, the papers deformed and it did not slide down when the desk inclined. Stabilize to make it look normal. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/198.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young South Asian man with curly hair sits in the passenger seat of a modern car with black leather seats and a panoramic sunroof. He wears a navy blue hoodie. Sunlight streams through the glass, highlighting his hair. He gestures animatedly while speaking, moving his hands in wide arcs. Through the partially opened car window and door, a serene beach and sparkling blue ocean are visible beyond a fence. The camera remains steady, capturing his expressive gestures and the bright, coastal background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: He's not South Asian. The window is not partially open.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/199.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young light-skinned woman with straight brown hair is shown in close-up. She holds a makeup brush near her cheek, with her ear and hair visible on the right side of the frame. The background is softly lit and out of focus. The camera remains steady, focusing on her face and the brush..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Blurry foreground [Motion & Physical Consistency]: Deformed hand and his brows remains the same. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/200.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A tall East Asian man with short black hair stands in a bedroom with white walls and simple furniture. He wears a gray T-shirt and dark pants. He raises his right index finger toward a metallic comment box suspended in mid-air, his gaze fixed on the box. Soft blue light reflects off the box, illuminating his focused expression. The camera remains steady, capturing the man and the softly lit bedroom background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: The box is not suspended in mid-air it was hanging.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/201.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly South Asian man sits on a wooden bench in a serene park during golden hour. The camera focuses closely on his elbow, showing rough, cracked skin with pronounced lines. A second hand, slightly weathered, gently massages a glossy, amber-colored oil onto the surface. The soft sunlight glimmers against the oil, and the elbow's texture gradually improves. Muted greenery and blurred flowers frame the background, while soft text appears: 'Care You Can Feel.'.\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The video has a garbage caption. [Motion & Physical Consistency]: There is an issue with the subject movement as the arm of the subject was overbend and it appears unnatural. Also with the cause-effect as the oil was not evenly spread to his arm. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: No. Object Existence: No. Subject-Object Interaction: There is another visible hand seen. The object is not Care You Can Feel.'.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/202.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A cartoon character with bright clothes and exaggerated features gleefully whistles while skipping across a tense battlefield. The character's arms swing playfully as they move, and their eyes are wide with excitement. Around them, the background shows scattered debris, smoky skies, and distant silhouettes of soldiers. The camera stays focused on the character’s lively motion, highlighting the contrast with the chaotic environment..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Texture issues in the background. [Motion & Physical Consistency]: Good. [Prompt Alignment]: Good.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/203.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A white teenage boy with short brown hair sits at a kitchen table. He grins as he holds a spicy chicken wing in his right hand and dips it into a small bowl of ranch dressing, coating the crunchy surface with the creamy sauce. The background shows a bright kitchen with countertop appliances. The camera remains steady, focusing on the boy’s dipping motion and cheerful expression..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Good. [Prompt Alignment]: He holds the chicken in his left hand.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/204.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young East Asian woman with slender eyes and slightly high cheekbones stands in a living room. She wears a fitted, stylish top and skirt. Her whole body is visible as she faces the camera, speaking directly toward it. The background features a sofa and soft lighting. The camera stays steady, capturing her posture and expressions..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Texture on the hair. [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/205.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Caucasian woman holds a pet python, her hands steady as she lets the snake curl around her fingers..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Her face isn't visible to determine if she is Caucasian.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/206.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged Russian woman with dark hair sits at a small wooden table in a cozy cafe. She wears a casual blouse and trousers. Shelves with pastries and merchandise line the walls behind her. She opens her new silver laptop, types out sales numbers, checks inventory lists, and occasionally glances at stacked receipts. The camera gently pans left to reveal her workspace and the cafe’s interior as she continues entering business data on her laptop..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the subject interaction as the hand of the subject was seen having a deformation. [Prompt Alignment]: The subject does not opens her new silver laptop, types out sales numbers, checks inventory lists, and occasionally glances at stacked receipts.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/207.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black boy with a curious smile watches intently as he threads a small gold needle for the first time, seated on the steps of a rustic porch. The needle gleams subtly as he carefully steadies the thread in his hands, while the afternoon sunlight filters through large maple leaves above..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Interaction with thread and needle is not in accordance to physical law. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/208.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A woman with dark brown skin and braided hair stands on a cliffside golf course overlooking turquoise ocean waves. She smoothly swings her golf club, striking the ball in a controlled practice shot. The background shows the ocean and rocky cliffs under bright daylight. The camera remains steady, capturing her focused movement and the scenic view..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: No [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: No - It looks like it was generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/209.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: Beneath the flickering streetlights on a rainy urban sidewalk at night, a teenage Latina girl with caramel-toned skin stops to rub her elbow, visibly irritated by redness and patches of dryness. She pulls out a sleek bottle of translucent serum with a silver cap and sprays it lightly on her joint. The droplets sparkle as they settle, instantly smoothing her skin. The camera zooms in to reveal her relieved expression against the blurred cityscape. The overlay text says, 'Heal anytime, anywhere.'.\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: There is garbage text visible in the video. [Motion & Physical Consistency]: The movement of the woman is not natural as her hands are overly bent. There is also a subject-object interaction issue as the way he handled the spray is not natural. There is also a cause effect issue as the spray did not reflect on her arms. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: No. Object Existence: No. Subject-Object Interaction: The skin tone of the woman is not Latina. The streetlights were not flickering. The subject and object interaction is not accurate with the prompt. The movement of the woman is not natural as her hands are overly bent. There is also a subject-object interaction issue as the way he handled the spray is not natural. There is also a cause effect issue as the spray did not reflect on her arms.  I believe the video is AI-generated because of this discrepancy. .\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/210.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A burly, middle-aged Latino man with a bushy beard stands at a workbench in a well-lit woodshop. He wears orange safety goggles and carefully pushes a mahogany plank through a table saw, trimming its edge with steady hands. The overhead lights cast a warm glow on the scene. The camera remains steady, focusing on his precise movements and the woodworking process..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: The overall movements, finger deformation, and the cause-and-effect fail, as the piece of wood is cut without producing any visible scraps. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/211.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A muscular middle-aged man with dark tan skin and short salt-and-pepper hair stands on a sandy beach at sunset. He wears a fitted gray T-shirt and dark shorts. Holding a weathered wooden bat in both hands, he swings and hits a bright orange plastic baseball high into the air. The background shows gentle waves and the sky glowing with warm sunset colors. The camera remains steady, capturing the man’s swing and the ball rising against the horizon..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The body movement and the cause-effect of the ball are all unnatural.\nThe ball suddenly deformed. [Prompt Alignment]: He did not  swings and hits a bright orange plastic baseball high into the air.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/212.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young biracial man with curly hair and tan skin shucks oysters on the deck of a fishing boat floating gently on calm waters. He works methodically with his knife, occasionally wiping his brow with the back of his hand. Nearby, nets filled with fresh oysters lie waiting, and the horizon fades softly in morning mist..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: The subject's movement is stiff and looks robotic as the subject's movement is stiff, not smooth. There is deformity in the subject's hands. He sliced the oyster shell, but the shell wasn't actually cut. Stabilize to make it look normal. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/213.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: In a futuristic pod filled with glowing blue light, an elderly East Asian man examines his elbow with care. His deeply lined skin contrasts sharply with the rough, dry texture of his elbow. His hand gently rotates the elbow, showcasing the skin’s condition. Text overlays float holographically around the elbow reading 'If you have this,' softly illuminating the area. The pod’s reflective silver walls remain out of focus, enhancing the emphasis on his skin and interaction..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage symbols. [Motion & Physical Consistency]: The cause-effect fails because the amount of liquid on his arm suddenly increases. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: The text don't match. His hand did not rotate the elbow. The cause-effect fails because the amount of liquid on his arm suddenly increases..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/214.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young East Asian woman with dewy skin gently rubs a dab of cream into her cheeks using circular motions in front of a bathroom mirror..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The cream on her face appears inconsistent and did not spread accurately. [Prompt Alignment]: She's not East Asian. She's not in front of the mirror.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/215.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A line of teenage girls walks toward the camera one at a time, each pausing to showcase their outfit—starting with a Black girl in a cropped varsity jacket, followed by a white girl in a pleated skirt and sweater, a Southeast Asian girl in wide-leg jeans and a graphic tee, and a Latina girl in a floral dress with boots..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: There is garbage text visible in the t-shirts. [Motion & Physical Consistency]: The movement of the girls is not natural as their facial features and body parts are deformed. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: The girl in wide leg jeans is not South Asian. The movement of the girls is not natural as their facial features and body parts are deformed. I believe the video is AI-generated because of this discrepancy. .\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/216.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged White man with a slim build and glasses sits at his office desk in a modern workspace. He wears a checkered long-sleeve shirt. Large windows behind him show buildings in an urban setting. He looks directly at the camera, maintaining steady eye contact and a calm, professional expression. The room is quiet and empty, with no other people present. The camera remains fixed at eye level, capturing his upper body and the office background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: pixelated shirt [Motion & Physical Consistency]: good [Prompt Alignment]: subject looks average build rather than slim build.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/217.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A close-up shot shows a floating mouth with glossy lips and bright white teeth centered in a neon-lit gaming room. The camera remains steady, highlighting the mouth against the glowing background. Several text bubbles float around the mouth, displaying humorous fresh thoughts. The shot is framed in a 9:16 aspect ratio, with a shallow depth of field and a wide-angle lens used..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The video has a garbage caption. [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/218.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young woman stands beside a grooming table in a pet grooming salon, holding a large orange Maine Coon cat. She gently places the cat on the table, where it lies with its front paws tucked under its body. Using a metal comb, the woman brushes the cat’s tail with smooth, gentle strokes, occasionally adjusting her grip to keep the cat steady. The background shows typical grooming salon equipment. The camera remains steady, focusing on the grooming process..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The hand appears to be deformed on the very first second and also on the last part of the video that it appears the hand follow the woman when she move to the left making it deformed. [Prompt Alignment]: There are two subjects. She did not place the cat on the table.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/219.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The subject, a Hispanic man, irons his shirt in front of the television, the iron gliding across the fabric with ease as he prepares for the day..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Blurry foreground, make it normal. [Motion & Physical Consistency]: Subject's movement and cause effect are unnatural. Make it normal. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: I think it's generated by AI because of the unnatural movement..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/220.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Kazakh man with short dark hair sits at a small café table. He wears a blue t-shirt and jeans. A plate with a burger and fries is on the table in front of him. The camera shows a medium shot of the man as he picks up the burger with both hands, takes a large bite, chews thoughtfully, and wipes his mouth with a napkin. The camera then slowly pans to the right as he reaches for a cup of soda, sips it, and sets it down. The background includes the café interior and soft lighting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: texture issues, garbage symbol [Motion & Physical Consistency]: hands are deformed. he picks up his burger to eat it but another burger appears behind it and no bite mark was seen after he took a bite [Prompt Alignment]: he did not sip the soda and set it down. he picks up the burger along with the napkin. There are two burgers on the plate instead of one\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/221.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A smooth slow zoom-in highlights the detailed stitching and texture of the leather shoe, emphasizing its craftsmanship and premium quality..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/222.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged man with tan skin and a rugged appearance stands in a dimly lit warehouse beside a sleek, stainless-steel bagging machine. He quickly places a stack of small paper bags under the machine’s dispenser. The machine releases measured amounts of fine white powder into each bag. The camera remains steady, capturing the man’s actions and the industrial setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Yes [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: Yes - It looks real and not generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/223.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The subject, a Hispanic woman, points to a video tutorial on her tablet, ready to learn how to bake a new recipe..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The movement subject with the object appear unnatural, as the actions seem forced and lack fluidity, disrupting the overall realism of the scene. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: No. Subject-Object Interaction: The subject does not appear to be Hispanic. There was  not video tutorial in the tablet. The movement subject with the object appear unnatural, as the actions seem forced and lack fluidity, disrupting the overall realism of the scene. I believe the video is AI-generated because of this discrepancy..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/224.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The subject, a Hispanic woman, washes the silverware, the water running as she cleans the forks and spoons..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: NULL [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/225.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A mature East Asian man with a slim build and short black hair sits at a wooden table in a cozy living room. He wears a dark t-shirt. Soft lighting and beige curtains create a warm backdrop. Facing the camera, he speaks with a calm, attentive posture, his facial expression naturally showing anticipation. His hands remain clasped on the table. The camera is fixed at eye level, capturing his steady gaze and subtle movements..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: textural issue on lips [Motion & Physical Consistency]: good [Prompt Alignment]: good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/226.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged Filipino man with short black hair sits in a mall, wearing a pale yellow polo-shirt. He holds a chip near his mouth with one hand, while a yellow bag of chips is partially visible out of frame. He brings the chip to his mouth, bites down, and chews with a crisp sound. After finishing the chip, he looks upward and relaxes his posture. The camera remains steady, capturing his actions and the mall background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage symbols [Motion & Physical Consistency]: Good [Prompt Alignment]: He is not finishing the chip but he already looks upward and relaxes his posture.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/227.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Black man breathes fire in a darkened space, the flames flickering brightly as they shoot from his mouth..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Cause-Effect failed. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/228.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: Deep underwater, a young man with light brown skin floats in a dreamlike scene surrounded by softly glowing bioluminescent creatures. He gently massages a futuristic metallic neck device that emits pulsating lights, while bubbles drift upward around him. The background is dark and surreal, illuminated by the creatures and the device. The camera remains steady, capturing the man’s relaxed expression and the ethereal underwater atmosphere..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Cause-effect failure as the bubble is not accurate. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/229.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Black man in his twenties stands in a brightly lit living room. He blows a bubble using a plastic wand, watching as the bubble floats in the air for a moment before it bursts. The background shows a sofa and a coffee table. The camera remains steady, capturing the man’s actions and the floating bubble..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Yes [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: No - It looks like it was generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/230.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A plate rests on a table in a brightly lit dining area. Several sushi pieces fall from the sky, landing one by one onto the plate. The camera remains steady, capturing the falling sushi and the plate below..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/231.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A late 20-year-old White man is shown from the neck up in a close-up, wearing a dark suit jacket with a crisp collar. he has a slighty chubby face, black hair, a little bit of facial hair, black eyes. He slowly turns his face to the left, then to the right, keeping his eyes wide and eyebrows slightly raised in a bewildered expression. The softly lit, blurred background reveals an out-of-focus forest. The camera remains steady, emphasizing his subtle facial movements and the details of his suit collar. He does not speak throughout the shot and mouth doesn't move. look left and right.\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/232.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Middle Eastern woman uses a disinfecting wipe to clean her bathroom sink after toothpaste has been splattered. The camera zooms in on the mess before the wipe, then shows the sink sparkling clean after the wipe has been used..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is a cause effect issue as the toothpaste was not removed after wiping. [Prompt Alignment]: The subject can't be determined due to lack of view. The camera did not zooms in on the mess before the wipe as the hand was already wiping.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/233.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young East Asian woman steps back onto a simple bridge under a nighttime sky. Her posture curves slightly, framing the silhouette of her waist. The fabric of her outfit falls naturally, and her relaxed arms form subtle lines leading to her poised hands. Her serene gaze faces the camera, and she blows a kiss softly into the air..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The movement of the woman is not natural as her legs are deforming. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: The woman did not blow a kiss. The movement of the woman is not natural as her legs are deforming.  I believe the video is AI-generated because of this discrepancy. .\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/234.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Black man with short hair walks through Times Square in a sharp navy blue shirt and black pants. The camera captures him in a full-body shot, highlighting his confident stride as he blends into the hustle and bustle of the city's vibrant atmosphere..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: There is deformity on the faces and body in the background, making it unnatural. [Prompt Alignment]: Good.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/235.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman holds a small gift box, smiling as she prepares to open it and see what's inside..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Interaction with the object as it deforms is not in accordance to physical law. [Prompt Alignment]: Good.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/236.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The subject, a South Asian man, jumps on the trampoline, his feet lifting off the surface as he reaches for the sky..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: NULL [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/237.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Hispanic woman stands in a well-lit living room. She raises her right fist and pumps it in the air in a celebratory gesture, smiling with satisfaction. The background shows a sofa and a coffee table. The camera remains steady, capturing her upper body and expressive movement..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Yes [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No - The video does contain nudity or harmful content. Cause-Effect: No - It looks like it was generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/238.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Hispanic woman in a white dress walks across the hot desert sand in her purple Crocs. The camera follows her lower body, focusing on how the Crocs grip the ground, giving her stability as she walks across the vast, arid landscape..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the subject movement as the feet was having stiffness. Also with the subject interaction as the hand was having deformation. Lastly with the cause-effect as no visible footprints was seen. [Prompt Alignment]: The subject is not clearly visible. The camera does not capture full body shot.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/239.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Caucasian woman dyes her hair blonde, the creamy color spreading through her strands as she works quickly..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The wordings in the video are just garbage symbols, not English. [Motion & Physical Consistency]: The subject's movement and the interaction between the subject and object are unnatural. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: I think it's generated by AI\n.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/240.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A metal spatula glides smoothly through a thick layer of white whipped cream, creating clean, even trails in the cream. The background is a vibrant solid color. The camera remains steady, focusing closely on the spatula’s movement and the texture of the whipped cream..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: good [Motion & Physical Consistency]: good [Prompt Alignment]: good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/241.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: Under the soft glow of artificial neon lights, delicate hands with caramel skin tones and nails painted in matte navy blue are seen against a polished black countertop. The person holds a transparent green bottle with 'Serum+ Glow' etched in metallic white. As the flip-top cap clicks open, a viscous, glowing gel with a faint multicolor shimmer emerges. A faint reflection of the neon colors dances on the surface of the gel as more is gently squeezed out. In the background, colorful geometric shapes are faintly blurred, creating an abstract, futuristic vibe..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The wordings in the product are just garbage, not English. [Motion & Physical Consistency]: Unnatural way of opening the product and it already spilled once its open. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: There is no  'Serum+ Glow' in the product..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/242.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged male restaurant worker in a black uniform walks from the back of a brightly lit dining area toward the camera, moving between tables and chairs. He approaches a visible yellowish liquid spill on the tiled floor near the camera and bends slightly, focusing on the spill. The background shows neatly arranged tables and chairs, with soft overhead lighting. The camera remains steady, capturing his movement and the dining area..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: The movement is not smooth and looks stiff. There is also a stain or spill on the floor that suddenly appears on the bottom part. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/243.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly Black man with a neatly trimmed gray beard stands on a wooden pier over a calm lake at sunset. He wears a flowing blue tunic that catches the soft pastel light. He performs a series of slow, precise tai chi movements, extending his arms gracefully and shifting his weight from one foot to the other. Gentle ripples in the water mirror his motions. The camera remains steady, capturing the peaceful lakeside setting and the man’s focused practice..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Yes [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: No - It looks like it was generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/244.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged man with tan skin and a thick beard stands in a desolate desert under a cloudy sky. He holds a pair of blue shears in his right hand and carefully trims the wool from a sheep whose coat appears faintly illuminated. The background shows sandy terrain and sparse vegetation. The camera remains steady, capturing the man’s focused movements and the unique desert setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The passing through of the scissor to the back of the sheep. [Prompt Alignment]: There are two pair of shears and he is holding it both hands.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/245.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young White man with short brown hair kneels on a beige carpet in a brightly lit living room. He leans forward and blows air onto the carpet with his mouth, directing his breath toward the surface. The background shows a sofa and a coffee table under natural daylight. The camera remains steady, focusing on the man’s actions and the carpet area..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The air coming out is inconsistent and appears too much or unrealistic. [Prompt Alignment]: He's not kneeling on the carpet in the video.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/246.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A South Asian man with medium brown skin applies a cooling under-eye gel using a metal applicator, focusing on his slightly puffy lower lids..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The wordings are garbage symbols. [Motion & Physical Consistency]: Interaction with object to object is not in accordance to physical law. Cause-effect failure as the gel increases its size. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: The subject is not South Asian..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/247.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged Russian woman with dark hair sits at a small wooden table in a cozy cafe. She wears a casual blouse and trousers. Shelves with pastries and merchandise line the walls behind her. She opens her new silver laptop, types out sales numbers, checks inventory lists, and occasionally glances at stacked receipts. The camera gently pans left to reveal her workspace and the cafe’s interior as she continues entering business data on her laptop..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the subject interaction as the hand was seen having a deformation. Lastly with the cause-effect as the subject was typing on the keyboard but no visible changes or reaction was seen on the screen display of the laptop. [Prompt Alignment]: The subject is not  Russian. The subject does not glance on the stacks of receipt.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/248.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young White woman with a fit build sits confidently in a vibrant gaming stadium, wearing a pixel print hoodie and dark jeans. She faces the camera with steady eye contact, speaking calmly and naturally as if having a conversation with the viewer. Her movements are minimal and controlled, with subtle gestures that convey confidence. The background shows softly glowing LED screens and esports-themed decor, creating an immersive atmosphere. The camera remains fixed, capturing her poised presence and the dynamic setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: good [Motion & Physical Consistency]: hand deformity during movement [Prompt Alignment]: good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/249.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A man and a woman sit side by side on a sandy beach, facing the ocean during a sunset. The sky is painted with vibrant orange, pink, and purple hues. The man holds a cheeseburger in his right hand, while the woman holds a plate of crispy French fries in her left hand. The camera remains steady, capturing both subjects and the sunset scenery..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The texture of the waves is flickering. [Motion & Physical Consistency]: The mouth deformed. [Prompt Alignment]: The woman is not facing the ocean\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/250.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Caucasian woman with straight shoulder-length hair sits at a vanity in a softly lit bedroom. She wears a light-colored top. Holding a concealer stick in her right hand, she applies it directly to blemishes on her face. She then uses her fingertips to blend the concealer with quick tapping motions. The camera remains steady, focusing on her face and hands as she completes the makeup step..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the cause-effect as the product was already being glide but still it does not leaves any marks upon application making it appear as unnatural as it does not leaves any traces or marks on her face. [Prompt Alignment]: The subject does not uses her fingertips to blend the concealer with quick tapping motions.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/251.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman plays a video game with a controller in hand, her fingers swiftly pressing the buttons as she navigates through the levels..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: NULL [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/252.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman with coily hair stands under running water, viewed from behind as she massages shampoo into her scalp, creating thick, foamy lather..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The hand movement and the cause and effect of the water from the hair are unnatural. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: No. Object Existence: No. Subject-Object Interaction: There is no running water and the subject's face in the video is not visible, so we cannot identify whether she is a young Black woman..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/253.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Caucasian man opens the door and exits his truck, stepping onto the gravel driveway, his boots crunching the ground as he heads toward the field..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Slight flickering issue. [Motion & Physical Consistency]: His left hand holding the door handle deforms unnaturally. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/254.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A teenage East Asian girl with a ponytail is pushing a small wooden wheelbarrow along a dirt path in a village, the cart carrying vibrant, hand-painted ceramic pots as she smiles against a bright blue sky..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Good [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/255.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged woman with olive skin and long flowing hair bench presses a sleek chrome barbell in the middle of a moonlit rock canyon. The surreal outdoor setting combines the rugged natural textures of stone with the stark silvery sheen of the equipment, as the milky way stretches across the night sky above her..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Good [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/256.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A muscular Arabic man in a fitted long-sleeve shirt and dark jeans stands indoors against a neutral background, with a large black clock and a dark wooden shelf displaying small decorative items behind him. He faces the camera, maintaining steady eye contact and speaking directly to the viewer. His facial expressions remain calm and professional throughout. The camera is fixed at eye level, capturing the casual home or office setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Texture on the beard and t-shirt [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/257.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A tall Black man walks along a quiet street in a classic black suit. The camera captures his full-body shot, showcasing the impeccable tailoring of the suit as he walks with purpose, his suit fitting perfectly with each step..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: His right hand has deformities. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/258.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An athletic young East Asian man with a sun-kissed tan balances on a vibrant turquoise surfboard, riding a towering ocean wave as the golden light of sunset reflects on the water's surface, creating a dreamy, cinematic backdrop..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: The subject's movement is stiff and looks robotic. Stabilize to make it look normal. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/259.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: On the edge of a rooftop under a dimly lit sky, a young man wields a hammer. He smashes a smartphone resting on a surface. The screen fractures in a cascade of fragments that dissolve slowly into the air..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: There is stiffness in the hands. Interaction with objects is not in accordance to physical law. Cause-effect failure as the phone is still intact. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/260.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A woman with freckles and auburn hair, wearing a yellow raincoat, leans over in a misty forest campsite to check the tire of her packed SUV. She grips a bright orange tire gauge, the device standing out against the damp, earthy colors of the forest floor and the gray tires..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Good. [Prompt Alignment]: Good.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/261.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged man with a beard and a denim jacket opens a water bottle at a beach campfire. The camera focuses on his hand as he twists the cap, showing the cool, refreshing water inside as the sound of waves crashes in the background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the subject movement as the hand of the subject was having deformation as it moves. Also with the subject object interaction as the object was seen deforming as the subject tries to open. Lastly with the cause-effect as the subject does not opens a water bottle at a beach campfire and it was inconsistent as it dispenses suddenly a liquid to the lid of the object. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: No. Object Existence: No. Subject-Object Interaction: The subject is elderly .The object is not water bottle. The subject does not  opens a water bottle at a beach campfire..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/262.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Caucasian man takes a sip from his coffee mug, letting the warmth and aroma surround him..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Good .\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/263.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The subject, a Hispanic woman, plays the keyboard with precision, adding depth to the band’s performance on stage..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: NULL [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/264.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: In a cozy living room, a South Asian woman with medium-length hair applies mascara to her lower eyelashes. The camera focuses on her eyes as she moves the wand in small, precise strokes, defining her lashes for a natural, enhanced look..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The hand movement and the cause and effect of the mascara are unnatural. [Prompt Alignment]: The subject in the video is not a South Asian woman.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/265.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Caucasian man plays the drums with enthusiasm, his fingers tapping rapidly on the hi-hat as the band prepares for the chorus..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Good [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/266.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Latina woman in her 30s with dark hair pulled back wears a loose T-shirt and sweatpants in a bright kitchen. She holds a spray can in her right hand and sprays thick white foam directly onto a greasy stovetop, watching as the foam expands and spreads over the stained surface. The background shows kitchen counters and appliances under natural daylight. The camera remains steady, focusing on her actions and the stovetop..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage symbols [Motion & Physical Consistency]: The cause-effect of the spray is unnatural. [Prompt Alignment]: The subject's face is not visible.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/267.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black man finishes recording and gently lowers his phone camera onto a table, the screen still facing up as he steps out of frame..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: There are deformities of object interaction and the subject's body features. Cause-Effect fails: Unnatural and Unrealistic of phone features and movement. Make it natural. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/268.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged Arabic man wearing a traditional white headscarf and a black robe sits in a modern, minimalistic living room with a television mounted on the wall. He faces the camera, maintaining steady eye contact. His facial expression shows subtle dissatisfaction, with slight tension around his mouth and a serious look in his eyes. He speaks calmly and firmly, expressing discontent through his tone and natural facial cues. His hand gestures are minimal. The camera remains fixed, capturing the calm and professional atmosphere..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: textural issues in beard [Motion & Physical Consistency]: good [Prompt Alignment]: good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/269.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Hispanic man in a tailored suit stands in front of a modern office building, striking a powerful pose. His arms are crossed, and he leans slightly forward, showing off the sharp lines and elegant design of his suit..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: The subject is not Hispanic.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/270.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The hand gently holds the blue textured cleaning brush, smoothly swaying it to highlight its surface in close-up..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Blurred and texture issue on the brush [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/271.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The camera slowly and steadily zooms in on the electric scrubber as it rests inside the bathtub, highlighting its design and placement against the clean white surface..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/272.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Singaporean Chinese woman with a fair complexion and glowing skin is shown in a close-up shot. She looks directly at the camera with a calm, confident expression. Her eyes are bright, and her lips are gently curved into a subtle smile. Her skin appears dewy and well-hydrated, with a natural sheen on her cheeks and high points. The background is softly blurred in warm neutral tones, emphasizing her face. The camera remains steady throughout the sequence..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/273.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black man stands in front of a large window overlooking a cityscape at sunrise, wearing a crisp white shirt and navy blazer. He holds a tablet in his right hand and gestures with his left hand as he addresses an unseen audience. Muted newsroom activity is visible in the background. The camera remains steady, capturing his confident posture and the early morning light streaming through the window..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/274.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman ties the laces of her sneakers in the park, the bright morning sunlight reflecting off the pavement as she prepares for her run..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: NULL [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/275.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman with short curly hair stands in a sunlit bathroom. She uses the fingertips of her right hand to gently spread a thick white cream across her cheek and forehead, blending it into her skin. The camera remains steady, capturing her smooth motions and the bathroom’s softly lit background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the cause-effect as the cream was not evenly applies to  the face and even though it was touch but it was not spread accordingly. [Prompt Alignment]: The subject was using her left hand to  gently spread a thick white cream across her cheek and forehead,\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/276.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A clear glass mug is gently placed under the spout of a drip coffee maker on a kitchen counter. As the mug settles into position, dark coffee begins to drip from the machine into the mug, gradually filling it and deepening the color of the liquid. The background shows a clean kitchen with soft morning light. The camera remains steady, focusing on the mug and the coffee as it pours..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: The mug was not being placed.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/277.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young man stands beside a bed in a softly lit bedroom. He holds a bundle of charging cables in one hand and quickly flings them onto the bed. The cords unravel in midair and land on the crumpled duvet. The background shows a nightstand and a window with closed curtains. The camera remains steady, capturing the motion of the cables as they settle..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is a subject-object interaction issue as the cable was merging and deforming on the hand of the  man. [Prompt Alignment]: The cord did not  unravel in midair.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/278.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Black woman in a casual short-sleeve top stands in a bright kitchen. She uses a large serving spatula to scoop up a steaming slice of lasagna from a glass baking dish, lifting it so that stretchy melted cheese trails behind. The background shows a countertop with kitchen utensils and warm lighting. The camera remains steady, focusing on her hands and the lasagna as she serves the slice..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: She did not serve the slice.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/279.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Hispanic man in full combat gear stands outdoors, wearing a helmet and tactical vest. He tightens the shoulder straps on his vest, then adjusts his helmet with both hands. The background shows a military base with vehicles and equipment under bright daylight. The camera remains steady, capturing his focused movements as he prepares for patrol..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Good [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/280.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged Indonesian man stands in a brightly lit urban area, holding a smartphone in both hands. He uses his right thumb to tap and swipe on the screen, navigating through a messaging app. The background shows city buildings and passing pedestrians. The camera remains steady, focusing on the man’s hands and the phone..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage text. [Motion & Physical Consistency]: Good [Prompt Alignment]: His face isn't visible to determine if he is Indonesian. He didn't use his thumb to tap and swipe.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/281.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: On an outdoor court, a Middle Eastern man in a yellow jersey takes a step-back shot from the baseline. The camera captures the ball soaring through the air, narrowly missing the hoop and hitting the board with a loud clang, bouncing off into the crowd..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage [Motion & Physical Consistency]: The movement is not smooth and looks stiff. The ball passes through the rim. [Prompt Alignment]: He's not Middle Eastern. He did not do a step-back shot.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/282.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A close-up shot shows an adult light-skinned man's hands playing an electronic keyboard indoors. He places his fingers on the keys and uses both hands to press different combinations of keys, playing various chords and notes in sequence. His fingers and hands shift smoothly along the keyboard as he continues to play. The camera remains steady, focusing on his hands and the keyboard..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Deformed fingers are unnatural. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/283.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly East Asian man with salt-and-pepper hair tied back stands in a minimalist room with warm track lighting. He maintains a serene expression and raises one eyebrow slightly. A piercer carefully aligns a thin needle to his upper ear cartilage and inserts a delicate titanium stud. The camera remains steady, capturing the man’s calm demeanor and the soft glow reflecting off the stud..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Not his upper ear cartilage.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/284.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The subject, a South Asian woman, digs in the garden, the fresh soil soft beneath her hands as she plants seeds..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: The subject's fingers are deformed and the movement is unnatural. The soil and seeds disappeared. Stabilize to make it look natural. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/285.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman with short curly hair stands in a bright kitchen, wearing a light blue T-shirt. She holds a small bowl of freshly grated Parmesan cheese in her left hand and uses her right hand to sprinkle the cheese evenly over a bowl of pasta in front of her. The background shows kitchen counters and utensils. The camera remains steady, capturing her as she adds the finishing touch to the pasta..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Cause-effect failure as the volume of cheese in her hand is not accurate that lands on the pasta. [Prompt Alignment]: She didn't hold a small bowl of freshly grated Parmesan cheese in her left hand.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/286.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young East Asian woman with straight black hair stands on a simple bridge at night, illuminated by soft ambient light. She wears a flowing dress that drapes naturally around her waist. Her posture curves gently, accentuating her silhouette, with her arms relaxed at her sides. She looks calmly at the camera and lifts one hand to blow a kiss into the air. The camera remains steady, capturing her graceful movement and the quiet nighttime setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: blurry hands. [Motion & Physical Consistency]: subject hands movement looks rigid. [Prompt Alignment]: subject did not lift one hand to blow a kiss into the air.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/287.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young East Asian woman with dewy skin gently rubs a dab of cream into her cheeks using circular motions in front of a bathroom mirror..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage symbols, not English. [Motion & Physical Consistency]: Unnatural hand movement and the cause and effect after spreading the cream are unnatural. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: She did not rubs the cream using circular motions..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/288.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The hand gently holding the hanging air freshener sways slightly, bringing the bottle into focus against the car dashboard and the view outside the windshield..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage symbol [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/289.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly East Asian man wearing a simple white shirt sits on a wooden bench in a lush garden at dusk, carefully lifting noodles from a patterned blue bowl as crickets chirp softly in the background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The way the insects fly and move is not natural. The cause-effect fails because the noodles seems not decreasing although he already keep on lifting and eating it. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: The animal is not a crickets. The way the insects fly and move is not natural. The cause-effect fails because the noodles seems not decreasing although he already keep on lifting and eating it..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/290.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young white woman with long, damp hair stands in a softly backlit shower. She raises both arms and uses her hands to lather shampoo at the crown of her head, spreading suds through her hair. The background shows a blurred shower wall with gentle light filtering in. The camera remains steady, focusing on her upper body and the motion of her hands as she works the shampoo into her hair..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: She is not a white woman.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/291.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Latino man stands upright in a modern academic research room, shown from the waist up with the camera steady at eye level. He wears a light blue button-up shirt with a medium neckline and covered shoulders, presenting a calm and composed expression. His hands are relaxed and remain out of frame. He maintains steady eye contact and speaks directly to the camera in a clear, confident manner. The background features a light wooden desk with sticky notes and a ballpoint pen, a grey fabric chair, an open wooden bookshelf with neatly arranged books, a round organizer holding scissors and paper clips, and a small cactus plant. Natural, balanced lighting fills the room. The scene is quiet and focused, with no sudden movements or background people..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/292.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged woman with dark brown skin and braided hair stands in a modern workshop, facing a chalkboard wall. She uses her right hand to sketch a chunky sports watch in bold strokes on the chalkboard. Sunlight streams through large windows, illuminating her steady hand as she draws. The background shows workbenches, tools, and prototypes arranged around the room. The camera remains steady, capturing her focused action and the bright, creative workspace..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The cause-effect of the marks in the pen is delayed. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/293.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Middle Eastern man in his thirties with short black hair stands beside a neatly made bed in a softly lit bedroom. He holds a smartphone in his right hand, lowers it onto the bed, then walks away from the camera. As he moves, the camera view slowly tilts upward toward the ceiling, capturing the room’s details and the man’s movement..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: good [Prompt Alignment]: good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/294.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Black man sits in a well-lit bedroom, wearing a casual T-shirt. He uses both hands to braid his own hair, carefully weaving the strands together into smooth, neat sections. The background shows a bed and a dresser. The camera remains steady, focusing on his hands and hair as he works..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: No [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: No - It looks like it was generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/295.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Middle Eastern woman with long dark hair stands at a vanity in a softly lit bedroom. She uses a dropper to apply golden-toned foundation drops across her face, then blends them in with quick, confident strokes using her fingertips. After blending, she leans closer to the mirror to show her even skin. The camera remains steady, capturing her actions and the cozy bedroom background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The two droppers merged. The foundation did not spread or mark accurately. [Prompt Alignment]: She did not apply it across her face.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/296.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Caucasian man irons his work shirt, the heat from the iron smoothing out the fabric as he works quickly..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: NULL [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/297.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A close-up of the wrist and forearm of a middle-aged Black woman, showcasing slightly uneven skin tone with faint wrinkles. The camera captures the hand of a young girl gently pressing the woman’s skin, highlighting its texture under the soft, golden glow of evening sunlight. A clean wooden table and a blurred garden in the background add warmth to the scene. Text overlays appear on-screen, reading 'caring starts here,' drawing attention to the nurturing touch and the skin’s condition..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage symbols. [Motion & Physical Consistency]: Good [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: No. Subject-Object Interaction: The subject's face is not visible, can't determine the age. The text does not match. The young girl did not press but only touch the woman.  Garbage symbols..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/298.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Black woman drops a small amount of serum onto her cheeks, massaging it into her skin with smooth, deliberate motions..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The movement of the woman is not natural. There is also a cause effect issue. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: The movement of the woman is not natural. There is also a cause effect issue.  I believe the video is AI-generated because of this discrepancy. .\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/299.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Black man looks at his phone, his face lighting up with excitement as he receives a notification on the screen..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: NULL [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/300.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The subject, a Hispanic woman, pours quinoa into a bowl, ensuring it spreads evenly across the surface for a healthy meal..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Good [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/301.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The subject, a South Asian woman, cleans the gutters on her home, the rain falling gently as she works to prevent clogs..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: NULL [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/302.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Southeast Asian woman stands in front of a camera, holding a smartphone in her right hand. She faces the camera and speaks, gesturing with her left hand as she talks. The background shows a softly lit indoor setting with neutral walls. The camera remains steady, capturing her upper body and expressive movements..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The video has a garbage caption. [Motion & Physical Consistency]: There is an issue with the subject interaction as the hand was having a deformation. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/303.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly South Asian man wearing a crisp white linen shirt and tan trousers walks out of a traditional open-air marketplace during golden hour. He carries several colorful woven shopping totes on his arms, with handcrafted jewelry visible through the bags. As he strides toward the entrance, the vibrant fabric stalls and market vendors recede in the background. Warm sunlight casts long shadows and highlights the details of his bags. The camera remains steady, capturing his movement and the lively market setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/304.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Black woman with rich skin tone dots under-eye cream along her orbital bone and blends it in with light tapping motions under soft bathroom lighting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is deformation in the subject's hand and the cream comes out unnaturally.\nThere is an object stick in her hand. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: There is an object in her hand and she did not blends the eye cream..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/305.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young East Asian woman with slender eyes and slightly prominent cheekbones sits on a sofa, wearing a white casual top and fitted black jeans. She looks toward a distant camera and speaks calmly, her posture relaxed. The background features a softly lit living room with neutral walls and a coffee table nearby. The camera remains steady, capturing her expression and natural gestures as she talks..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/306.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: In a room, a teenage boy tosses clothes onto a bed. The garments land in a loose pile as faint light from outside casts soft shadows in the background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The movement of the man's hands appears unnatural, as they look stiff and deformed. There is also an object-object interaction issue, as the clothes he is tossing are unnaturally deforming. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: The movement of the man's hands appears unnatural, as they look stiff and deformed. There is also an object-object interaction issue, as the clothes he is tossing are unnaturally deforming. I believe the video is AI-generated because of this discrepancy. .\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/307.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young South Asian woman with tied-back hair carefully stacks two delivery boxes on a table before organizing the contents inside..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The upper part of the second box below appears to move on its own. [Prompt Alignment]: She's not South Asian. The content is not visible. She's not stacking two delivery boxes.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: No. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/308.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly Black woman with gray-streaked braids tied back neatly sits in the passenger seat of a car. She wears a green knit cardigan. She gestures and talks animatedly, waving her hand and then pointing firmly, her gold-painted nails catching the light as she moves. The car interior is softly lit, with the dashboard and window visible in the background. The camera remains steady, focusing on her expressive movements and lively conversation..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Her finger passing through the other fingers. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/309.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The subject, a Hispanic woman, blows the colorful autumn leaves off the driveway, the leaves spiraling into the air before falling neatly..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: NULL [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/310.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged white woman with fair skin and shoulder-length blonde hair stands in a sunlit room. She holds out her left hand, showing a row of lipstick swatches—nude pink, classic red, plum, and burnt orange—on the back of her hand. The camera remains steady, focusing on her hand and the swatches, with soft natural light filling the background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Deformed fingers. [Prompt Alignment]: There are more than four swatches in the video.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/311.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A South Asian woman with long dark hair holds her phone in one hand, taking a close-up photo of a flower. She steadies her hand and gently adjusts the phone’s position to focus on the flower. The background shows a garden with greenery and soft natural light. The camera remains steady, capturing her careful movements and the flower in detail..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Good. [Prompt Alignment]: Not South Asian. No flower visible. She didn't hold her phone in one hand, and didn't take a close-up photo of a flower.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/312.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A South Asian woman in business attire opens the front door of her car and sits in the driver’s seat. She adjusts the rearview mirror with one hand, then presses the start button to turn on the engine. The background shows a residential driveway in the morning light. The camera remains steady, capturing her actions inside the car..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Her hand deforms unnaturally. She didn't turn the car keys but the car turned on which is unnatural. [Prompt Alignment]: She isn't South Asian. There is no start button but only a keyhole with a key on it. She didn't adjust the rear view mirror.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: No. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/313.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young South Asian woman stands indoors, holding a pile of newly bought clothes pressed against her chest with both arms. She smiles brightly and looks directly into the camera. The background shows a softly lit living room with a sofa and decorative cushions. The camera remains steady, capturing her cheerful expression and the clothes in her embrace..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/314.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young boy with caramel skin and curly hair sits on a wooden porch at sunset. He uses both hands to wrap a soft, elastic bandage around his scraped knee. Warm golden light casts gentle shadows on the porch, and the background is filled with the soft sound of crickets. The camera remains steady, capturing the boy’s focused movements and the peaceful evening setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the subject interaction as the hand was having deformation and passing through to the object he was controlling. [Prompt Alignment]: The subject does not wrap a soft, elastic bandage around his scraped knee. Warm golden light casts gentle shadows on the porch, and the background is filled with the soft sound of crickets.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/315.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Middle Eastern woman with golden olive skin drops a clear serum onto her cheekbone and gently blends it in with her fingertips under warm bathroom lighting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: The subject's hand deformed and the movement is unnatural. The product disappeared. Stabilize to make it look natural. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/316.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Southeast Asian man wearing a winter jacket and helmet rides a futuristic snowmobile across a landscape of glowing, multicolored snow. He glides smoothly forward, leaving a trail behind him as his scarf flares outward. Auroras dance in the sky above, casting shifting lights on the snow. The camera follows him from the side, capturing his motion and the vibrant, otherworldly scenery..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the cause-effect as the snow ashes was not seen on the back of the tires and no visible tire marks was seen and the reaction of the snow was inconsistent being seen as it was seen only on the front view not in the back. [Prompt Alignment]: The subject is not clearly visible.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/317.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A teenage boy with fair skin and freckled cheeks lounges in a top-of-the-line gaming chair, his headphones with a flexible microphone glowing faintly with integrated LED lights. The room is filled with vivid neon hues, reflecting off his glasses as he leans forward, gripping a controller, ready to face the challenges on screen in a high-stakes cyber environment..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The wordings are garbage symbols. [Motion & Physical Consistency]: Interaction with the headphone microphone is not in accordance to physical law. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: I think it's generated by AI\n.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/318.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Singaporean Chinese woman with straight black hair lies on a bed in a softly lit bedroom. She gently blinks her eyes open, revealing a calm, slightly smiling expression. The background features a warm brown-to-golden gradient, suggesting early morning light filtering through the curtains. The camera remains steady, capturing her relaxed posture and the gentle movement of her eyelids..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: texture issues [Motion & Physical Consistency]: good [Prompt Alignment]: she does not blink her eyes open.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/319.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A teenage Black girl with braided hair stands in a dimly lit garage, surrounded by scattered tools. She lifts a small silver hammer in her right hand and forcefully brings it down onto a sleek black smartphone lying on a wooden workbench. The phone’s screen shatters on impact, sending small glass fragments outward. The background shows shelves and tools in the shadows. The camera remains steady, capturing the action and the glint of broken glass..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is a cause effect issue as the phone was already cracking even though it was not hit yet and the cracks are inconsistent. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/320.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: In a futuristic workshop with sleek metallic walls, a middle-aged South Asian man wearing safety goggles and a silver cloak manipulates a robotic arm that shapes a glowing horseshoe, emitting a faint humming sound..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Good [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/321.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Southeast Asian teenage girl with a pink clay face mask stands in front of a bathroom mirror. She holds her hands under her chin and wiggles her fingers, striking a playful, mock-serious beauty pose. The background shows bathroom tiles and a countertop. The camera remains steady, capturing her reflection and expressive gesture..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Texture [Motion & Physical Consistency]: Good [Prompt Alignment]: She did not wiggles her fingers. There's no countertop.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: No. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/322.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young East Asian man dynamically launches off his red surfboard, executing an aerial trick mid-wave. The ocean backdrop is vivid and roaring, with sunlight beaming brightly across the frothy wave he's conquering..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: The way the surfboard shot upward appeared unrealistic. His hand passed through his head. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/323.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young woman with straight brown hair and a middle-aged man with short dark hair sit side by side in a bright beauty studio with clean white walls and soft lighting. They both wear comfortable T-shirts and jeans. A professional wearing gloves applies warm wax to the woman’s forearm using a flat spatula, then presses a cloth strip firmly over the wax. Next, the specialist quickly removes the strip in one fluid motion. The man watches the process, while the background displays beauty products neatly arranged on shelves. The camera remains steady, capturing the waxing treatment and the relaxed studio atmosphere..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the subject interaction as the hand of the person wearing gloves was having a deformation. Also with the cause-effect as the white product was not evenly applies and inconsistent being applies to the arm of the woman and it sudden disappear and moving  making it appear unnatural. [Prompt Alignment]: The therapist does not  quickly removes the strip in one fluid motion.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/324.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman with short curly hair sits on a chair in a well-lit music studio. She holds a bass guitar, positioning her left hand on the fretboard and her right hand near the strings. She plucks the strings with her right hand while her left hand shifts along the fretboard to play different notes. The camera remains steady, focusing on her hands and the instrument as she continues to play, with the sound of deep bass notes filling the background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: There is deformity in the hand. [Prompt Alignment]: Good.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/325.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young White man with cool-toned fair skin, an oval face with a rounded jawline, and ice-blue almond eyes sits in an e-sports chair in a gaming room. He has delicate facial features including light gold arched eyebrows, a small rounded nose, full M-shaped lips, and smooth skin without stubble. His long, waist-length light blonde hair is straight and neatly styled. He wears a burgundy satin shirt and fitted dark gray pants, presenting a slender figure. The video shows his upper body and arms, facing the camera as he speaks with a slightly indifferent expression and subtle makeup. He occasionally moves his hands while holding a smartphone. The camera remains steady, capturing the conservative style and gaming room background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Deformed teeth [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/326.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Black man in his 30s stands on a sturdy ladder beside his house. He reaches up with one hand to scoop leaves and dirt from the gutter, dropping the debris to the ground below. The background shows the house’s siding and a few trees under daylight. The camera remains steady, capturing his actions and the area around the gutter..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Yes [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: No - It looks like it was generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/327.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman with short curly hair stands in a bright kitchen, wearing a casual T-shirt. She holds a bottle of soy sauce in her right hand and pours it directly over a plate of barbeque-glazed ribs on the counter. The background shows kitchen cabinets and utensils. The camera remains steady, focusing on her action and the food..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage text is visible. [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/328.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A tall Middle Eastern man with a trimmed beard stands inside an empty metal wheelbarrow at a dimly lit construction site. He grips the barrow’s handles with both hands and carefully maneuvers it forward, steering between shadows cast by cranes and construction equipment. The overhead light casts uneven illumination, highlighting the man’s focused movements and the surrounding dark environment. The camera remains steady, capturing his progress through the site..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The eye movement of the subject is unnatural. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/329.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Caucasian man sits on a bench in a park, his legs crossed to display his classic canvas sneakers. He wears a casual hoodie and joggers, which pair well with the shoes. The background shows green grass and trees under natural daylight. The camera remains steady, focusing on the man and his outfit..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/330.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Asian woman floats in zero gravity inside a dimly lit space, steadying a box and opening it to reveal a glowing object. Her face lights up with awe as she reaches for it..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the subject interaction as the hand was having a deformation and it was seen passing through to the object. [Prompt Alignment]: The subject was not opening  box it to reveal a glowing object. Her face lights up with awe as she reaches for it.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/331.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Middle Eastern woman with long dark hair sits at a wooden table in a cozy home living room. She holds a small white smartphone box in both hands, lifts the lid to reveal the device inside, and carefully removes the smartphone. She examines the phone closely, then peels off the protective film. She arranges the phone and accessories neatly on the table. The camera remains steady, capturing her actions and the relaxed home setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Blurred around the hand. Texture issue in the lip [Motion & Physical Consistency]: Deformed hand and box lid [Prompt Alignment]: There is no protective film and accessories. She did not remove the smartphone and did not peel off the protective film.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/332.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman with curly hair and a cozy yellow scarf stands on a bustling sidewalk at the edge of a vibrant street market. The camera smoothly moves along the line of people as she leans slightly forward, carefully selecting a bouquet of pink-wrapped flowers from a nearby cart. Her fingers brush past the other bouquets, and the soft sunlight highlights her smiling face. The background hums with chatter and movement, filled with cheerful decorations and urban textures that capture the lively energy of the market..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The wordings in the background are just garbage. [Motion & Physical Consistency]: The hand movement of the subject and the body feature in the background are unnatural. And the way her hands go through the wrapper are unrealistic. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: The video is generated by AI..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/333.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: POV from inside a shopping bag: A Black woman’s hand with glossy nude-painted nails reaches into the bag and grasps a boxed skincare product. The background shows the soft interior of the shopping bag, with diffused light filtering in. The camera remains steady, focusing on the hand and the product as it is picked up..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage symbol [Motion & Physical Consistency]: Good [Prompt Alignment]: Cannot determine if it's a woman. She did not reach into the bag and grasps a boxed skincare product as she already held it\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/334.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged Latina woman with warm-toned skin and hoop earrings stands in a sunlit room. She tilts her head slightly and presses a soft strip of peach-colored wax onto her eyebrow with one hand, while glancing at her phone in her other hand for instructions. The background shows gentle morning light streaming through a window. The camera remains steady, focusing on her face and hands as she completes the action..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The effects of the wax strip on her eyebrows aren't realistic as it should have removed her brow hair realistically. [Prompt Alignment]: The wax strip didn't really press on her eyebrows but on her eyelids instead.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/335.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Black man slices an orange with ease, the vibrant color contrasting against the cutting board as he prepares a snack..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The first slice seems to have doubled. The slice, the mark, and the other orange slice moving on its own is not accurate and inconsistent. [Prompt Alignment]: The face is not visible.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/336.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly East Asian man with a fit build stands in a gym under soft neon lighting. He wears a gray T-shirt and black athletic pants. Holding a small blue kettlebell in each hand, he performs front raises by lifting both arms straight forward to shoulder height, then lowering them in a controlled motion. The background shows mirrors lining the walls and gym equipment. The camera remains steady, capturing his focused exercise and the gym setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: good [Motion & Physical Consistency]: good [Prompt Alignment]: good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/337.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Hispanic man in his late twenties wearing a gray athletic T-shirt stands in a gym with workout equipment in the background. He holds a plastic energy drink bottle in his right hand, lifts it to his mouth, and takes a sip while looking ahead. The camera remains steady, capturing his action and the gym setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage text. [Motion & Physical Consistency]: Good [Prompt Alignment]: He looks white rather than a Hispanic.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/338.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The person gently adjusts the pillows on the beige sectional sofa, subtly shifting position to create a tidy and inviting seating area..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the subject interaction as the hand of the subject was having a deformation. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/339.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The subject, a Hispanic man, uses a grooming comb to brush his cat’s short fur, the cat rolling over in delight..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage symbols. [Motion & Physical Consistency]: The subject's movement and the interaction between the subject and subject is unnatural. There is a cause-effect failure. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: The subject's movement and the interaction between the subject and subject is unnatural. There is a cause-effect failure..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/340.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Latina opens a travel-size toothpaste and places it into a geometric pouch next to cleanser jars in a bedroom..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Movement is stiff and not smooth. The bottle slightly changed shape and pockets suddenly broke when he placed the bottle on it. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/341.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young boy with deep brown skin and curly hair sits cross-legged in a cozy living room. He plays a bright green recorder, surrounded by books and soft, warm lighting from a nearby lamp..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage symbols [Motion & Physical Consistency]: The hand movement is not natural and there is a deformation in the object while she touches it. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/342.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Middle Eastern man responds to a sidewalk interview in downtown, gripping the microphone and gesturing animatedly while his group listens and cheers him on..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The wordings are garbage symbols. [Motion & Physical Consistency]: There is deformity on the hand and face, making it unnatural. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: He didn't grip the microphone..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/343.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Middle Eastern woman in a red cap and uniform opens a fresh pizza box with a smile, revealing a cheesy margherita pizza under the warm kitchen lights..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage text. [Motion & Physical Consistency]: Her hand movement as well as interaction with the pizza  and the carton of pizza deforms unnaturally. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: The woman didn't open the box of pizza..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/344.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A muscular Arabic man with short dark hair stands indoors in front of a white door and a white tiled wall, resembling a gym locker room. He wears a fitted black T-shirt and athletic pants. He faces the camera, speaking directly while maintaining steady eye contact. His facial expressions remain neutral with subtle hints of dissatisfaction. The lighting is bright and even, clearly illuminating his face and upper body. The camera is fixed at eye level, capturing his engaged and direct demeanor..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Texture on beard [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/345.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Hispanic woman with long dark hair stands on a softly lit stage, wearing a flowing dress and elegant ballet flats. She gently shifts her weight onto one foot, drawing attention to her shoes. The background features subtle decorative elements, and the camera remains steady, focusing on her poised stance and the details of her outfit..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: There is deformity in the legs as only one leg is visible. [Prompt Alignment]: The subject is not clearly visible. The object is not ballet flats. She didn't gently shift her weight onto one foot.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: No. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/346.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: Five people stand together on a city sidewalk at night, illuminated by streetlights. They form a semi-circle facing a man on the right, who holds up his phone toward the group as if recording or showing something. The man in the center and the woman beside him both rest their hands on their hips. The woman on the left holds her phone in front of her, glancing at its screen. The group remains in place, appearing to interact with the man holding the phone. The camera stays steady, capturing the group’s body language and the nighttime city background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The mouth movement, the cause effect of the phone and the deformation in the hand are all unnatural. [Prompt Alignment]: There is another phone that the man in black holds.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/347.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged Black man with a salt-and-pepper beard stands beside a frozen lake in the middle of winter. He launches a neon blue disc golf Frisbee, its vivid color contrasting with the pale snow and the glittering, icy trees in the background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The video has a garbage caption. [Motion & Physical Consistency]: Good [Prompt Alignment]: The subject does not  launches a neon blue disc golf Frisbee\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/348.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An older woman with gray hair and a sunhat opens a bottle of water in a botanical garden. The camera zooms in on her hand as she twists the cap off, with the clear water ready to refresh her as she enjoys the surrounding flowers..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: NULL [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/349.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A metal spatula glides smoothly through a thick layer of white whipped cream, creating clean, even trails in the cream. The background is a vibrant solid color. The camera remains steady, focusing closely on the spatula’s movement and the texture of the whipped cream..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the cause-effect, as the creamy product was inconsistently spread; even when it was glided by the object, it was still not coated firmly, making the application appear uneven and unrealistic. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/350.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Middle Eastern woman with long dark hair sits at a vanity in a softly lit bedroom. She tilts her head slightly and uses her right hand to apply mascara to her upper lashes, carefully brushing from the roots to the tips. Her dark lashes fan out as she completes the motion. The camera remains steady, focusing closely on her face and the glossy finish of her lashes, with a blurred background showing a mirror and decorative items..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Good. [Prompt Alignment]: She didn't use her right hand to apply mascara to her upper lashes.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/351.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A towering Black woman in athletic gear takes a deep breath in a remote desert setting as she throws a lightweight matte-black throwing axe at a dead tree. The arid surroundings are bathed in the orange glow of sunrise, sand swirling lightly with her movements..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Slight flickering issue. [Motion & Physical Consistency]: Her overall movements are unnatural. The effects of her swinging the axe are unrealistic as well. [Prompt Alignment]: It isn't a lightweight throwing axe and she didn't throw it.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/352.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Arabic woman wearing a traditional beige dress with intricate embroidery and a white headscarf stands in a warmly lit reception room decorated with festive lanterns. She faces the camera, maintaining steady eye contact and speaking with a soft, relaxed expression that subtly reflects relief. Her facial expressions remain natural and comfortable throughout the sequence. The camera stays fixed, capturing her presentation and the festive background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Texture on the dress. [Motion & Physical Consistency]: Good [Prompt Alignment]: The woman is not Arabic.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/353.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A woman with dark hair stands in a room with white walls and a ceiling light fixture. She faces the camera directly, her mouth wide open and her eyes wide in a state of surprise. The background remains unchanged throughout the video. The camera stays steady, capturing her expression and subtle movements..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Texture on lips [Motion & Physical Consistency]: Good [Prompt Alignment]: There's a extra brush which is not mentioned in the prompt. The woman eyes is not wide in a state of surprise.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/354.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged white woman with rosy cheeks and a fur-lined hood bikes along a snowy farm road during golden hour, her mountain bike's thick tires leaving crisp tracks in the pristine landscape as the orange sunlight reflects off the snowdrifts..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the subject movement as it appears unnatural. Also with the cause-effect as no visible marks was seen upon riding the bike. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: There is an issue with the subject movement as it appears unnatural. Also with the cause-effect as no visible marks was seen upon riding the bike. .\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/355.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young light-skinned man skates forward on an outdoor ice rink at night in an urban setting. He bends his knees and swings his arms upward, then jumps into the air with both arms raised and legs bent. He lands on the ice with one leg extended behind him and continues gliding forward in a crouched position. The background features city lights and buildings. The camera follows his movement smoothly from the side..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: The body movement is unnatural, and the body parts seem to interchange. There are no traces in some parts of the ice where he's there. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/356.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Hispanic teenage boy with jet-black hair tied in a bun walks confidently out of a vibrant video game store. He wears a casual hoodie, ripped jeans, and neon sneakers. In both hands, he carries two bold red shopping bags. The background shows shelves filled with colorful game boxes and glowing monitors inside the store. The camera remains steady, capturing his exit and the lively store interior behind him..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage symbols. [Motion & Physical Consistency]: Good [Prompt Alignment]: did not walk out of the store.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/357.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Arabic man stands confidently in an open-style art studio, shown from the waist up with the camera steady at eye level. He wears a light blue button-up shirt with a medium neckline and covered shoulders. His posture is upright and composed, with his hands relaxed and out of frame. He looks directly at the camera and speaks, maintaining steady eye contact and a calm demeanor. The background features a standing easel with a canvas, a worktable with paint stains, a ceramic jar of paintbrushes, a wooden tray with sculpture tools, and a pot of green trailing plants against the wall. Natural, balanced lighting highlights the artistic setting. The camera remains steady, focusing on the man as he speaks, with no sudden movements or background people..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: The subject is not Arabic.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/358.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Turkish woman with long dark hair sits at a café table. She holds a small cup of coffee in her right hand and takes a sip while looking ahead. After a moment, she spits the coffee out in shock, her face showing a startled and surprised expression. The background features a cozy café interior with soft lighting. The camera remains steady, capturing her actions and the café setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Deformed hand and the cause-effect of the coffee is unnatural. [Prompt Alignment]: The table is not visible and she did not spits the coffee out in shock, her face showing a startled and surprised expression.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/359.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Black man wades through the crystal-clear water of a stream, stepping over rocks as he enjoys the cool sensation of the water..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/360.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An extreme close-up shot of matcha green tea ice cream shows its smooth, dense surface. Delicate spoon marks are visible across the top, and a faint powdery sheen of matcha is present. The background is softly blurred, focusing attention on the texture and color of the ice cream. The camera remains steady throughout the sequence..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/361.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Caucasian woman with shoulder-length brown hair sits at a kitchen table. She holds a large slice of watermelon in both hands, takes a bite, and chews as juice drips down her chin. The background shows a bright kitchen with fruit on the counter. The camera remains steady, capturing her eating the watermelon..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Yes [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: Yes - It looks real and not generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/362.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A white teenage boy grins as he dunks a spicy wing into ranch dressing, the creamy sauce spreading over the crunchy surface..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: The hand movement and the cause and effect of the chicken after he dips it are all unnatural. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/363.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Brazilian man lies on a simple mattress in a softly lit room. He wears a loose t-shirt and shorts, holding a white pillow close to his chest with his legs curled. A digital alarm clock sits on the floor beside the mattress. The camera starts with a medium shot of the man sleeping, then slowly pans right to reveal the clock. It returns to a steady shot of the man as he shifts slightly in his sleep..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage symbol / Blurred around the face and arm [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/364.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: Inside a minimalist, modern loft at night, a young white woman with freckles and auburn hair sits at a small wooden table. She holds a felting needle in her right hand and uses it to sculpt a tiny blue bird from wool on a wooden mat. A soft, warm desk lamp illuminates her focused face. The camera remains steady, capturing her careful needlework and the cozy loft setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/365.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Middle Eastern woman with long dark hair sits at a vanity in a softly lit bedroom. She smiles gently as she uses her right hand to apply warm-toned blush to her cheeks, moving her fingers in small circular motions. Her left hand rests on the table. The background shows a mirror, makeup items, and a lamp. The camera remains steady, focusing on her face and hand movements..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is a cause effect issue as the cheek of the woman did not have visible changes. [Prompt Alignment]: The woman is using her left hand to apply the blush.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/366.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: Inside a bustling night marketplace, a middle-aged South Asian woman wearing a vibrant sari stands beneath lantern light. She holds a small box with intricate patterns in both hands and carefully unwraps it. She lifts the lid to reveal a slim, gold-trimmed smartwatch inside, which reflects the flickering lantern light. Her smile widens as she looks at the watch. The camera remains steady, capturing her reaction and the lively market background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The deformed hand is unnatural. [Prompt Alignment]: She is not unwrapping.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/367.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Black man flips a bottle in the living room, the bottle spinning gracefully before he catches it in one motion..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: The subject's movement appears to be stiff. Make it smooth. There are deformities of subject-object interaction. Make it more natural. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/368.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The subject, a Caucasian father, holds a baby bottle filled with milk and feeds his baby, watching with affection as the baby sucks contently..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Slight flickering issue. [Motion & Physical Consistency]: The overall movement of the baby is unnatural. The milk from the bottle didn't lessen realistically. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/369.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: zoom up of grass of football field - drone shot.\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Good. [Prompt Alignment]: Good.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/370.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An asado de 100 metros is a large outdoor barbecue event. The video shows a long row of grilled meats on metal racks, with several adults standing nearby watching. The background includes picnic tables and greenery under daylight. The camera remains steady, capturing the arrangement of food and the atmosphere of the event..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the subject interaction as the hand of the subject was having a deformation. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/371.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A muscular middle-aged Black man with short hair stands in a dimly lit industrial gym. He wears a sleeveless athletic shirt and dark shorts. He delivers a series of rapid, rhythmic punches to a red leather punching bag, causing it to swing back and forth with each impact. Beads of sweat glisten on his skin under the light from a single overhead bulb. The camera remains steady, capturing his focused movements and the industrial gym setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: His hands passed through. Cause-effect fails as he doesn't hit the hanging punching bag but moves. [Prompt Alignment]: Good.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/372.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young White woman with straight light brown hair sits at a vanity in a softly lit bedroom. She uses a dense makeup brush to apply pale foundation in small dots across her face, then gently blends the foundation with the brush in circular motions, evening out her skin tone as she works. The camera remains steady, focusing on her face and hands as she blends. The background shows a mirror and decorative items on the vanity..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the cause-effect as the foundation was not evenly applies to the face even if the subject was gliding on the brush but still it was not applies accordingly to the face. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/373.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Texan cowboy with hat opens a packet of cookies while sitting on a picnic blanket. The camera focuses on his hand as he tears open the packet, revealing the chocolate-dipped treats inside, ready to share with friends..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The wordings in the product are just garbage, not English. [Motion & Physical Consistency]: The hand movement is not natural and chips suddenly disappear. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: He did not  opens a packet of cookies while sitting on a picnic blanket..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/374.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young South Asian woman with wet, shoulder-length black hair sits indoors. A hand holds a hair treatment container in front of her..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage symbols [Motion & Physical Consistency]: Deformed hand and the cause-effect of the white substance floating are seem unnatural. [Prompt Alignment]: We cannot assume that she is sitting since the only visible is her face only.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/375.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: On an outdoor basketball court, a Caucasian man wearing a blue jersey stands facing a South Asian woman in a yellow jersey. He holds a basketball in both hands and makes a chest pass toward her. The woman stands ready with her arms outstretched to receive the ball. The background shows the court surface and a chain-link fence under daylight. The camera follows the basketball as it moves from the man to the woman, capturing both players and their actions..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Yes [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: No - It looks like it was generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/376.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Latino man with a slim build stands in front of a green screen showing a lush green landscape with sheep grazing. He wears a blue shirt. The setting appears calm and serene. He faces the camera, maintaining steady eye contact while speaking with a relaxed, friendly expression. The camera remains fixed, capturing his natural and confident presentation style..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Texture on lips and hair. [Motion & Physical Consistency]: Good [Prompt Alignment]: The man look more to medium build than a slim build.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/377.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: On the edge of a vast desert under a bright turquoise sky, a young boy with freckles and sandy blond hair sits on a large rock. He focuses on playing a slender, wooden recorder that harmonizes with the wind carrying grains of sand..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: NULL [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/378.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young East Asian man with a strong athletic build and short hair sits at a desk in a modern room with green lighting in the background. He wears a fitted casual t-shirt. Facing the camera, he speaks steadily while maintaining eye contact and occasionally using natural hand gestures. His expression remains calm and focused as he shares fitness or lifestyle insights. The camera stays fixed, capturing the relaxed atmosphere and the subtle green-lit ambiance..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: texture issues in lips [Motion & Physical Consistency]: good [Prompt Alignment]: good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/379.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A South Asian man with a beard pours fresh mint body wash into his hand in a tranquil forest. The camera zooms in on his hand as he dispenses the cool, green body wash, focusing on the smooth texture as it sits in his palm..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: The subject is not fully visible.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/380.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Middle Eastern man in a gray hoodie stands at a kitchen counter. He uses a metal spatula to lift a large square of lasagna from the corner of a baking tray, the golden crust crackling slightly as he serves himself. The background shows a bright kitchen with countertops and utensils. The camera remains steady, focusing on his hands and the lasagna..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is a an object interaction issue as the spatula was merged on the baking tray. [Prompt Alignment]: The subject can't be determined due to lack of view.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/381.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A South Asian man stands in a well-lit bathroom. He holds a comb in his right hand and runs it smoothly through his hair several times, creating a neat and polished appearance. The background shows a mirror and bathroom fixtures. The camera remains steady, focusing on his grooming action..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: No [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: No - It looks like it was generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/382.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A South Asian woman with straight hair uses the Maybelline Super Stay Brow Pencil to define her right eyebrow in a chic coffee shop. The camera zooms in on her face as she makes small, controlled strokes, giving her brows a defined and polished look..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: There is garbage text visible in the pencil. [Motion & Physical Consistency]: The woman's movement appears unnatural, as it is stiff. Additionally, there is a cause-and-effect issue, as the pencil marks appear inconsistently. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: No. Object Existence: No. Subject-Object Interaction: The woman is not South Asian. The pencil is not Maybelline Super Stay Brow Pencil. The woman's movement appears unnatural, as it is stiff. Additionally, there is a cause-and-effect issue, as the pencil marks appear inconsistently. I believe the video is AI-generated because of this discrepancy. .\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/383.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Middle Eastern woman gently scrubs conditioner through her shoulder-length hair, seen from behind as water trickles down her back in a steamy bathroom setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The hand movements aren't consistent and it deforms unnaturally. Water is pouring onto her hair yet it doesn't look wet in real. [Prompt Alignment]: Her face isn't visible to determine if she is Middle Eastern. There was no conditioner and she didn't rub any onto her hair.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: No. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/384.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A black man wearing a blue blazer and a pink shirt stands against a pink background. He has a red pin on his left lapel and is smiling at the camera. The camera remains steady, focusing on the man and the pink background. No visible text appears in the frame..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/385.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly South Asian man in a crisp white linen shirt and tan trousers walks out of a traditional open-air marketplace at golden hour. He carries several colorful woven shopping totes on his arms, with handcrafted jewelry visible inside the bags. The background shows vibrant fabric stalls and vendors bathed in warm sunlight, with long shadows cast across the market entrance. The camera remains steady, capturing his stride and the lively marketplace atmosphere as he moves toward the exit..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/386.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The subject, a Hispanic woman, puts on her sandals, the straps securing comfortably as she heads out for a day in the sun..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Interaction failure with the sandal. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/387.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The subject, a Black man, drags his rolling suitcase through a crowded train station, wearing a jacket and jeans as he moves quickly to catch his ride..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/388.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: Inside a bustling night marketplace, a middle-aged South Asian woman wearing a vibrant sari stands beneath lantern light. She holds a small box with intricate patterns in both hands and carefully unwraps it. She lifts the lid to reveal a slim, gold-trimmed smartwatch inside, which reflects the flickering lantern light. Her smile widens as she looks at the watch. The camera remains steady, capturing her reaction and the lively market background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Texture issue on the box and the clothes [Motion & Physical Consistency]: Deformed hand [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/389.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A teenage boy with dark curly hair and freckles sits in a green kayak, gripping a paddle with both hands. He maneuvers the kayak through foamy white rapids, pushing off the side of the boat to navigate around rocks as water splashes over his face and the surrounding river. The camera follows the boy from a low angle, capturing the dynamic motion of the kayak and the cascading water as he moves through the rapid..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: He keeps padling but he isn't moving. [Prompt Alignment]: He isn't navigating the rapids as he didn't move from his place.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/390.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A close-up shot through the glass of a washing machine shows bright red, blue, and green shirts mixed with faded denim tumbling gently inside. Water droplets bead and slide down the curved glass window, filling most of the frame. The camera slowly pans to follow the swirling fabric and wet texture, with soft white light rays creating movement and depth. The background is softly blurred, emphasizing the spinning clothes and water droplets..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Texture [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/391.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A close-up shows an adult hand lifting a slice of pepperoni pizza from the whole pie. As the slice is pulled away, golden melted cheese stretches in long strands between the slice and the remaining pizza. The background features a warmly lit kitchen countertop. The camera remains steady, focusing on the hand and the pizza..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The effects when he pulled a slice of pizza in the middle aren't realistic and natural. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/392.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged man with olive-toned skin and a scruffy salt-and-pepper beard balances on a bike in the middle of a desert, the cracked, sun-baked ground stretching endlessly around him as heatwaves ripple in the golden light..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The mark on the sand is inaccurate and the way it keeps moving for no reason is a cause-effect failure. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/393.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young adult stands on a beach in Costa Rica, holding a large empanada in both hands. They jump onto a wave, riding it toward the shore. The background shows palm trees, clear water, and a city skyline. The camera follows the subject’s movement as they surf the wave..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The texture of the subject is flickering. [Motion & Physical Consistency]: The overall movement, the deformation, and the cause-and-effect of sudden appearance of the board all seem unnatural. [Prompt Alignment]: He is  holding a large empanada in onehand.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/394.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Hispanic man in his 30s with short black hair stands in a well-lit kitchen. He holds a plastic beverage bottle in one hand and shakes it vigorously for several seconds to mix the contents. He then lifts the bottle to his mouth and takes a sip. The camera remains steady, capturing his actions and the kitchen background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The beverage did not spill or overflow although it had no lid or cap. [Prompt Alignment]: The video does not show him sipping it because the content did not reach his mouth.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/395.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Caucasian man stands in a bright kitchen, holding a fresh orange in his left hand and a manual juicer in his right. He presses the orange half onto the juicer and squeezes, extracting juice that flows into a clear glass below. Sunlight streams through the window, illuminating the countertop. The camera remains steady, focusing on his hands and the juicing process..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Yes [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: No - It looks like it was generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/396.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A close-up shot shows an adult hand holding a black pen above a lined notebook. The pen tip briefly hovers over the paper, then the hand writes the letters “xi” in smooth, cursive script to complete a word on the line. After finishing the letters, the pen pauses and lifts slightly from the page. The background is softly lit, with the notebook filling most of the frame. The camera remains steady, focusing on the hand and writing..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: The hand movement and the traces of the pen comes out unnaturally. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/397.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Caucasian woman enjoys a sundae, her spoon dipping into the creamy ice cream as she savors the layers of toppings..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: There is an issue with the cause-effect, as the subject merely touched the spoon to the ice cream, yet it suddenly started melting. This appeared unrealistic and disconnected from the actual interaction. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/398.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young East Asian woman sits indoors, facing the camera. She lifts both hands to adjust her glasses near her temples, then runs her fingers through her hair, letting it fall naturally around her shoulders. She relaxes her arms and occasionally tilts her head or shifts her posture. The background shows a softly lit room with neutral decor. The camera remains steady, capturing her subtle movements..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Good [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/399.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Caucasian man in a blue basketball jersey stands at the free-throw line on an indoor court. He holds a basketball in both hands, bends his knees slightly, and shoots the ball with a smooth, straight motion toward the hoop. The camera follows the ball as it arcs through the air and passes cleanly through the net. The background shows blurred spectators in the stands..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage [Motion & Physical Consistency]: The ball passes through the rim. The ball moves inaccurately. [Prompt Alignment]: He's not standing in the free throw line.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/400.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The subject, a South Asian woman, chisels a wooden frame, smoothing out the edges as she prepares it for painting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: NULL [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/401.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman plays the cymbals during a concert, her movements sharp and synchronized with the rhythm of the band..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage symbols, not English. [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/402.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman pulls the lever on a slot machine, her eyes glued to the spinning reels as she waits for the results..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: NULL [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/403.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A medium shot shows an East Asian woman in a white shirt standing in a bright kitchen with pale wood cabinets and a clean countertop. She stands near the center island, watching several pieces of sushi as they arc through the air in front of her. Some sushi pieces are suspended mid-air, while others hover just above the counter. The camera slowly pans left as she reaches out and gently catches one of the pieces, then places it down beside the others. She continues to smile throughout the sequence..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: blur sushi [Motion & Physical Consistency]: sushi duplication as woman started with two sushi and it duplicated to 5. When the sushi hits the board some of them were fuzed into the exisint sushi. [Prompt Alignment]: Some sushi pieces were not suspended mid-air, while others did not hover just above the counter. Subject did not watch the sushi but was tossing them around the counter.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/404.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Asian man stands confidently in a modern academic research room, shown from the waist up with a steady, eye-level camera. He wears a light blue button-up shirt with a medium neckline and covered shoulders. He maintains a calm, focused expression and speaks directly to the camera, holding steady eye contact. His hands remain relaxed and out of frame, with no gestures or object interaction. The background features a light wooden desk with sticky notes and a ballpoint pen, a grey fabric chair, an open bookshelf with neatly arranged books, a round stationery organizer with scissors and paper clips, and a small cactus plant. Natural, balanced lighting highlights the serene, scholarly setting. No background people or visible text..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/405.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: In a peaceful garden, a Middle Eastern woman with long hair opens a rustic brown and green cake box. The camera captures her hand as she removes the lid, unveiling a decadent carrot cake with a generous layer of cream cheese frosting and chopped nuts..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Flickering issue [Motion & Physical Consistency]: The hand movement and the box opens unnaturally. [Prompt Alignment]: The subject's face is not visible.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/406.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A female aesthetician in a white lab coat stands beside a counter in a brightly lit training room, her hair pulled back neatly. She demonstrates the use of a professional high-frequency facial device on a realistic mannequin head, holding the glass electrode in her right hand as it glows gently. She gestures toward the electrode, explaining its features. Several adult students seated at the counter lean forward, watching her technique closely with focused expressions. The camera arcs behind her shoulder to frame both the device and the attentive students, capturing the clear, softly illuminated classroom setting with shelves of skincare equipment visible. The cinematic training moment is highlighted with crisp light and detailed action..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: texture issues in hair [Motion & Physical Consistency]: hands deformed when gesturing [Prompt Alignment]: good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/407.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman with neatly manicured fingers holds a pastel pink spatula. She uses the spatula to spread a thick, glossy white cream across a reflective glass surface, making slow, deliberate sweeping motions. Each pass creates gentle ripples in the cream, highlighting its smooth texture and uniform appearance. Soft daylight reflects off the surface, adding a warm glow. The background is minimalistic and clean. The camera films from above, focusing on the spatula’s movement and the cream’s texture..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Good. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/408.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Southeast Asian woman with straight black hair applies warm beige foundation to her cheeks, forehead, and chin in even drops. She uses a flat brush to blend the foundation smoothly across her face. She then looks at the camera and smiles. The background shows a well-lit vanity area with a mirror and beauty products. The camera remains steady, capturing her makeup application and expression..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The effects of the foundation when spread aren't natural and realistic. [Prompt Alignment]: She didn't blend the foundation.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/409.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Black teenage boy wearing a gray hoodie sits in a brightly lit room. He leans forward toward the camera, grinning, and waves with both hands in a playful manner as if on a video call. The background shows a desk and shelves. The camera remains steady, capturing his cheerful gesture and expression..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: There is deformity in the hand. [Prompt Alignment]: He's not teenage.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/410.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: Under a surreal lavender sky on an abstract, glowing field of crystal shards, a middle-aged East Asian man sits cross-legged, his arm resting atop iridescent crystals. The camera zooms in on his rough elbow skin as he sprinkles a shimmering liquid from a vial. The liquid’s golden tint highlights the transformation, nourished textures emerging under his touch. Light dances around him, giving subtle hints of magic, while the text 'Even here, skin needs care' subtly brightens along the edge of the screen..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage symbols, not English. [Motion & Physical Consistency]: The unnatural movement and the interaction with the other hand at the side appear unnatural. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: No. Object Existence: No. Subject-Object Interaction: There is another hand in the video and his arm is not resting atop iridescent crystals..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/411.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged man in a navy shirt stands in his home office. He uses both hands to lift his black desk to a standing position, then picks up a pen and begins writing in a notebook on the desk. The background shows bookshelves and a window with daylight streaming in. The camera remains steady, capturing his actions and the organized office setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: textxure issue and flickering of notebook's black binder [Motion & Physical Consistency]: good [Prompt Alignment]: Man is lowering his desk from a high standing postion to a lower position instead of lifting it\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/412.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: In a quiet forest, a South Asian man with a beard opens a can of lime soda. The camera focuses on his hand as he pulls the tab, capturing the fizzy bubbles rising to the top, with the fresh, clear soda inside..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: There is garbage text visible in the can. [Motion & Physical Consistency]: The movement of the man's hand appears unnatural and stiff. There's a subject-object interaction issue, as the can tab becomes deformed when pulled. Additionally, there's a cause-and-effect issue, as the soda inside the can is flowing without a clear trigger. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: There are two cans visible in the vide but was not mentioned in the prompt. The movement of the man's hand appears unnatural and stiff. There's a subject-object interaction issue, as the can tab becomes deformed when pulled. Additionally, there's a cause-and-effect issue, as the soda inside the can is flowing without a clear trigger. I believe the video is AI-generated because of this discrepancy..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/413.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: In a futuristic indoor arena with vibrant holographic displays, a tall woman with dark curly hair kicks a glowing blue soccer ball, her sleek boots skidding slightly on the iridescent metallic floor..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Her overall movements are unnatural. Interaction with the ball is unnatural as well as the ball moves without a cause. [Prompt Alignment]: She didn't kick the ball.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/414.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young East Asian teen in a sleek black and gold tai chi uniform practices deliberate stances and arm motions within a glowing energy-filled realm, with shimmering orbs of light flowing around them in harmony with their intricate movements..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Good [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/415.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A tall white man with long hair stands on a mountainside next to a bubbling stream, holding up a dripping tie-dyed scarf in shades of purple, pink, and orange against a backdrop of towering pines..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Flickering issue [Motion & Physical Consistency]: The body movement is not natural. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/416.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A South Asian woman stands in front of a large gong on a stage with other musicians nearby. She holds a mallet in her right hand and strikes the gong with controlled, rhythmic motions, coordinating her movements with the ensemble. The background shows stage lighting and musical instruments. The camera remains steady, focusing on her precise playing..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: The subject's movement is stiff and robotic, lacking smoothness, and their hands appear deformed. Stabilize to make it look normal. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/417.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: In a cozy bedroom, a South Asian man with a neatly trimmed beard uses a hair dryer to style his short hair. The camera focuses on his hand as he directs the dryer's airflow, with his hair flying slightly as he shapes it, achieving a fresh, styled look..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage [Motion & Physical Consistency]: The shirt also moves like it is being blower but the blow dryer wasn't aimed to the shirt. [Prompt Alignment]: He's not South Asian. He did not shape it.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/418.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly European woman with silver hair tied in a bun stands in a vintage wooden barrel on a misty hillside. She gently stomps grapes with her bare feet, holding onto the edge of the barrel with her weathered hands for balance. The background shows a softly glowing vineyard scattered across the hillside in the morning fog. The camera remains steady, capturing her movements and the serene landscape..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: blurry hair. [Motion & Physical Consistency]: good [Prompt Alignment]: No grape can be seen in the video. Subject did not gently stomps grapes with her bare feet.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/419.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A high-speed camera pans across the reflective surface of Cloud Gate, also known as The Bean, in Chicago on a sunny day. The camera moves quickly along the curved metal structure, capturing its shimmering reflections and the busy city background. The shot includes a crowd of people walking and standing near the sculpture, with colorful clothing and bright sunlight creating vivid contrasts. The camera remains in motion, highlighting the dynamic interplay of light and shadow on the sculpture..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Flickering issue [Motion & Physical Consistency]: Deformed faces. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/420.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young boy with dark curly hair and freckles stands on the porch of a rustic cabin. He holds a deck of vibrant blue cards in both hands and energetically shuffles them, moving his arms quickly. The background features tall pine trees and dappled sunlight on the wooden porch. The camera remains steady, capturing the boy’s actions and the natural setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The hand movement and the deformation in his hand is unnatural. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/421.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The hand gently holds the HealthPro product box, subtly adjusting its position to showcase the packaging design and details with a smooth, steady motion..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage text. [Motion & Physical Consistency]: Good [Prompt Alignment]: It is HealthPro but it isn't a box.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/422.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman with short curly hair sits at a cluttered wooden desk in a sunlit room. Her hand reaches out, gripping a tangled bundle of bright white, black, and red charging cables, their rubbery texture catching the light. The desk is scattered with a laptop, a coffee mug, and a small succulent plant, suggesting a busy workspace..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage symbols, not English. [Motion & Physical Consistency]: The overall movement, the interaction with the floating wire and the cause and effect of sudden changes are all unnatural. [Prompt Alignment]: She's already holding the charging cables. She did not reaches it.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/423.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Latino man stands confidently in a modern industrial workshop, shown from the waist up in a medium shot. He faces the camera directly, maintaining steady eye contact as he speaks clearly and authoritatively. He wears a dark blue work shirt with a mid-coverage neckline and covered shoulders, presenting a clean, professional appearance suitable for machinery manufacturing. The background is clean and slightly blurred, showing a metal workbench with surface scratches, a tool wall with hanging wrenches and screwdrivers, a red rolling toolbox, a tray of small assembly components, and a pair of orange safety gloves on the desk corner. The camera remains steady at his eye level as he addresses the viewer in a calm, steady posture without gestures..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Deformed object and sudden movement in the background. [Prompt Alignment]: The subject is not Latino and He did not maintain steady eye contact.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/424.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A tight close-up of a young Black woman’s lips as she applies a glossy nude lipstick, the shine intensifying with each pass of the applicator..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The marks of lipstick is inconsistent and did not spread evenly. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: The whole face is not visible. The marks of lipstick is inconsistent and did not spread evenly..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/425.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Black teenage girl with braided hair sits at a vanity in a warmly lit bedroom. She holds a dual-ended brow brush and uses it to shape and smooth her eyebrows, brushing them in short, gentle strokes. The camera remains steady, capturing her focused grooming and the cozy bedroom background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Deformed hand and the cause-effect of the brush is unnatural. [Prompt Alignment]: There is no vanity\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/426.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Black man in his thirties stands at a kitchen counter under bright lighting. He holds a red apple in his left hand and uses a small knife in his right hand to slice the apple into bite-sized pieces. The camera remains steady, showing the cutting action and the glossy surface of the apple on the countertop..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: The apple remains in its original form when sliced. Stabilize to make it look normal. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/427.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman with curly hair sits in a dimly lit café, her fingers moving across the screen of a smartphone. Subtle light reflects off her glasses as soft colors from a nearby sign cast gentle tones onto the walls around her..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Deformity on the smartphone, making it not in accordance to physical law. Cause-effect failure as her touch and the display is not accurate. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: I think it's generated by AI\n.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/428.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young East Asian man standing, then walking in front of a white picket fence and a white house, with greenery in the background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/429.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman with curly hair, wearing a light sweater, stands in a cozy living room. She holds a small gift box in both hands, smiling as she lifts the lid to open it and looks inside. The background shows a sofa and soft lighting. The camera remains steady, capturing her anticipation and the warm setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Deformed finger [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/430.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Southeast Asian man in his 30s with short black hair stands in a bright laundry room. He wears a light blue T-shirt and holds a neatly folded stack of freshly washed laundry in both hands, lifting it up and displaying it proudly to the camera with a cheerful smile. The background shows shelves with detergent and a washing machine. The camera remains steady, focusing on the man and the laundry..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: textural issue in hair [Motion & Physical Consistency]: good [Prompt Alignment]: good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/431.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A white teenage girl with straight blonde hair sits cross-legged on her bed in a softly lit bedroom. She wears a light blue T-shirt and shorts. Holding a bottle of creamy body lotion, she dispenses some onto her hands and gently rubs the lotion onto her legs in smooth, circular motions. The camera remains steady, showing her focused application and the cozy bedroom setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Her hand is deformed. Cause-effect fails as the amount of lotion being spread seems much and is inaccurate. [Prompt Alignment]: She didn't hold the bottle of lotion and didn't dispense some onto her hands.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/432.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: In a bustling city rooftop at twilight, a middle-aged South Asian man with a salt-and-pepper beard juggles flaming torches with precision. The urban skyline in the background sparkles with lights, and the fiery pattern contrasts with the cool blue-purple hues of the sky..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: The hand movement and the subject-object interaction fails because of the way he holds the flaming torches are all unnatural.\nThe sparkles of lights appears unnaturally. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/433.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A close-up shot shows a wooden desk with a matching drawer unit and a potted plant placed on top. The background features a modern interior with clean lines and neutral colors. The camera remains steady, focusing on the desk and its details..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/434.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A South Asian woman stands in a garden bed surrounded by greenery. She uses both hands to scoop and turn over fresh soil, revealing a layer of mulch and seeds beneath the surface. As she continues to mix the soil, a small plant with green leaves and purple flowers becomes visible. The camera remains steady, capturing her hands working in the soil and the gradual transformation of the garden bed..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The flower in the small plant suddenly becomes three and also it suddenly stands even if it was not planted well. [Prompt Alignment]: The subject is not fully visible.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/435.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young African American man with short hair stands on a sunny playground. He wears a fitted athletic T-shirt and shorts. He runs toward a colorful gymnastics block, then dives forward, tucking his head and rolling over the block with controlled motion. The background shows playground equipment and bright sunlight. The camera captures the sequence smoothly, focusing on his athletic movement..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: textural issues with hair [Motion & Physical Consistency]: body deformed at the end of video [Prompt Alignment]: there is no colourfyl gymnastics block, just what looks like a stack of colourful mats. he jumps pber and falls on his head\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/436.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young East Asian man of average build, wearing a white shirt and light blue jeans, sits comfortably on a striped armchair in front of a plain beige wall. A modern lamp with a white lampshade stands behind him on the right. He faces the camera, maintaining steady eye contact and speaking in a calm, conversational manner. The setting is relaxed and informal, with a fixed camera capturing his natural expressions and presentation style..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: blurry mouth and teeth during movement [Motion & Physical Consistency]: good [Prompt Alignment]: good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/437.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: In a sports hall, a young Black woman wearing a purple tank top and black athletic shorts dribbles a basketball while moving across the court. Her right hand controls the ball as she looks ahead, maintaining focus. The background shows the gym floor and basketball markings under bright indoor lighting. The camera follows her from the side, capturing her steady dribbling and movement..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Her movement was stiff and not smooth. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/438.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Caucasian woman and a Hispanic man walk side by side along a busy street. The woman holds a small microphone near her mouth with her right hand, while the man gestures with his left hand as he speaks. Both remain standing and continue their conversation as they walk. The background shows pedestrians and city buildings. The camera remains steady, capturing their interaction from the side..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The passing through of their feet are unnatural. [Prompt Alignment]: The man is holding also a microphone which is not mentioned in the prompt.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/439.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Black man with short hair adds sliced strawberries and bananas into a food processor on the kitchen countertop. The camera zooms in on the fresh fruit as it gets added, ready to be blended into a smoothie..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The wordings are garbage symbols. [Motion & Physical Consistency]: Interaction with fruits is not in accordance to physical law. Cause-effect failure as the on the bananas as it suddenly appears in his hand. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: I think it's generated by AI\n.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/440.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Black man hands a small gift box to his partner, watching their eyes light up with excitement as the gift is revealed..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: NULL [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/441.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A white man with fair skin and faint dark circles stands in a minimalist bedroom. He uses his fingertips to gently smooth a small amount of eye cream under both eyes. The background features a neatly made bed and simple decor. The camera remains steady, focusing on his face and hands as he applies the cream..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Texture issue on the skin and lip [Motion & Physical Consistency]: In the first 1, 2 secs, the cream did not change when the person spread it [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/442.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged man with olive skin and salt-and-pepper hair stands barefoot in a sunlit courtyard surrounded by wine-making tools and clay pots. He repeatedly steps down on a pile of grapes with his bare foot, pressing and trampling them in a steady, methodical motion. The ground is covered in crushed grape skins, and the air is filled with the earthy scent of the fruit. The camera remains steady, capturing his actions and the busy courtyard setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the subject interaction as the feet was having deformation and only four toes was seen. Lastly with the cause-effect as the subject was stamping the the grapes but no visible liquid was dispensing out even though the feet was stamping on it. [Prompt Alignment]: The subject is not clearly visible.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/443.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A teenage Middle Eastern boy with olive-toned skin leans against a graffiti-covered wall in an urban alleyway. He dots a lightweight, metallic green serum onto his face using a small dropper, tapping it briskly into his skin under the golden haze of the setting sun. The liquid seems to vanish instantly, leaving a matte finish..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: His movement on applying the serum is unnatural. Cause-effect failure on the serum as it did not dispense accurately. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: I think it's generated by AI\n.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/444.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: In a whimsical, dreamlike kitchen with pastel-colored cabinets, an excited teenage boy with freckles and curly auburn hair pulls a tray of perfectly baked cookies from the oven. The cookies emit soft steam, and the surrounding kitchen glows with a magical, golden hue..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The overall movements of the boy are unnatural. The oven suddenly opens by itself, then another tray of cookies disappears when the door of the oven rests on it. The boy is holding the tray that came from the oven without any hand mittens. The tray should be hot. [Prompt Alignment]: The face isn't fully visible to determine if he is a teenage boy. The oven is inaccurate since the door is small for its size.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: No. Object Existence: No. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/445.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The subject, a Hispanic man, pours beer from a bottle into a glass, the foam settling as he takes a sip..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/446.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Southeast Asian woman in a denim jacket responds to a street interview question, gripping the mic with a smile as her friends nod and wait their turn behind her..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: The left hand of the girl in the blue jacket deforms and moves unnaturally. The microphone floats. Stabilize to make it look normal. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/447.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Caucasian woman with blonde hair uses a disinfecting wipe to clean her kitchen table. The camera shows the before and after: the wipe removes food stains and crumbs, leaving the table spotless and shiny..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Blurry foreground, make it normal. [Motion & Physical Consistency]: The hand movement and the cause and effect of the crumbs are unnatural. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: The subject's face in the video is not visible, so we cannot identify whether she is a Caucasian woman with blonde hair and it did not leaves the table spotless and shiny..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/448.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black man with a fade haircut stands in a dimly lit room. He holds a white bandana in one hand and twirls it gently, then dips the edges of the bandana into a shallow dish filled with marbled green and orange dyes. The dyes swirl together as the fabric touches the liquid. The background is shadowy, with soft, moody lighting. The camera remains steady, focusing on the man’s hands and the dyeing process..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: His hands were deformed. [Prompt Alignment]: The shallow dish filled with marbled green and orange dyes is not visible. He didn't dip the edges of the bandana into a shallow dish filled with marbled green and orange dyes.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/449.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly Black man wearing weathered orange leggings stands on a makeshift sandy path beneath a dome-like sky over a vast desert landscape at twilight. He bends his knees and swings his arms back, then jumps forward three times in succession, each leap sending a spray of sand into the air. The sun dips low behind distant dunes, casting a warm amber glow that highlights the arc of his jumps. The camera remains steady, capturing the man’s movements and the shimmering sand against the glowing sky..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The effects of the sand are unrealistic. [Prompt Alignment]: He didn't jump forward, he was jumping in the same place.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/450.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young woman with medium skin tone and shoulder-length dark hair stands in a brightly lit store aisle lined with shelves of personal care products. The video begins with a close-up of a blue and white toothpaste tube held by a hand, displaying the front of the tube. The camera then cuts to the woman as she holds the same toothpaste tube in her right hand, turning it slightly to show different angles. She faces the camera and smiles, keeping the tube visible. The background features neatly arranged shelves and products under bright overhead lighting. The camera remains steady throughout the sequence..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage [Motion & Physical Consistency]: Good [Prompt Alignment]: She's holding the tube in her left hand in the next scene.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/451.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young East Asian man wearing a white apron flips a perfectly crisped chicken thigh with tongs over a pan of hot oil, the camera focusing on the rich, golden color of the frying chicken against a clean, modern kitchen backdrop..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: The oil drips unnaturally from the chicken. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/452.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged Black man with salt-and-pepper hair stands in front of a large mirror in a softly lit room, meticulously tying a crimson silk necktie with precision and care. The tie glimmers subtly under the warm, golden glow of the overhead light as he adjusts his shirt collar and takes a deep breath of satisfaction..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The movement of the man's hands appears stiff. There is also a subject-object interaction issue as the hand of the man went through his tie. [Prompt Alignment]: The man does not appear tying his necktie.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/453.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young White woman with straight brown hair stands at a table in a brightly lit room. She holds a white phone box with colorful graphics in both hands, then peels off the clear plastic wrapping. Next, she removes a white charging cable from the box and peels off its protective sticker. She then lifts the phone out of the box, holding it with both hands. The background shows a tidy living space, and the camera remains steady, capturing her actions from the front..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The plastic seems deformed and also the charging cable. The tear on the white object is not accurate and the plastic seems to be floating and did not fall directly. [Prompt Alignment]: She did not lift the phone out of the box.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/454.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Middle Eastern woman dabs a tiny amount of under-eye cream beneath each eye, then gently taps it in with her ring finger in front of a vanity mirror..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: There is garbage text visible in the video. [Motion & Physical Consistency]: The movement of the woman while applying the cream is not natural, as her hand appears stiff. There is also a cause-effect issue, as the cream is not evenly applied. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: The movement of the woman while applying the cream is not natural, as her hand appears stiff. There is also a cause-effect issue, as the cream is not evenly applied. I believe the video is AI-generated because of this discrepancy..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/455.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young South Asian man in a blue shirt stands at his kitchen counter. He holds a ceramic coffee pot in one hand and pours freshly brewed coffee into a white cup placed on the counter. The camera smoothly zooms in on the cup, focusing on the dark coffee filling the cup and the steam rising from its surface. The background shows a bright kitchen setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: His face isn't visible to determine if he is South Asian.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/456.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Caucasian man applies deodorant to his underarms, ensuring it glides on smoothly for a clean, fresh scent..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The overall movement, the interaction with the incorrectly applied deodorant, and the resulting cause and effect all appear unnatural. [Prompt Alignment]: The subject in the video is not a young Caucasian man and he is not applying to his underarms.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/457.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A white American male game streamer sits in an electronic gaming room. He holds a smartphone horizontally in his right hand, facing the screen as he plays a game. His face is turned away from the camera, with only his mouth and jawline visible. The background shows gaming equipment and dim lighting. The camera remains steady, focusing on the streamer’s upper body and the phone..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The video has a garbage caption. [Motion & Physical Consistency]: Good [Prompt Alignment]: The subject is not clearly visible to identify as  white American\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/458.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A white teenage boy lightly rubs a subtle pink blush onto his cheekbones while looking into a small mirror..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the cause-effect as no visible blush was seen upon application and no visible marks was seen as the subject was rubbing the face. [Prompt Alignment]: The subject does not rubs a subtle pink blush onto his cheekbones while looking into a small mirror.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/459.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman hands a gift box to her friend, smiling as she passes it over during a birthday celebration..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: NULL [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/460.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Korean woman with clear skin and straight black hair stands in a softly lit bathroom. She wears a light gray T-shirt. She pulls a fluffy pink headband over her hairline, then gently adjusts it with both hands to keep her hair back. The background shows a bathroom mirror and sink. The camera remains steady, focusing on her face and upper body as she finishes adjusting the headband..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Good. [Prompt Alignment]: She didn't pull a fluffy pink headband over her hairline.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/461.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The subject, a Hispanic woman, completes a ballet routine with a stunning arabesque, holding the pose as she finishes her performance..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: She is not a Hispanic and it's not a stunning arabesque.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/462.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A close-up captures a white woman’s lips slightly parted as she swipes on a coral glossy lipstick, creating a smooth, glass-like finish..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Blurry foreground, make it normal. [Motion & Physical Consistency]: The hand movement and the cause and effect after swiping the glossy lipstick are unnatural. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: I think it's generated by AI because of the unnatural movement..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/463.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: At a sunny outdoor cafe, a young Black woman with braids opens a blue and white chips packet. The camera captures her hand as she tears the top of the packet, revealing the golden, crispy chips inside, ready to snack on while enjoying her coffee..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The video has a garbage caption. [Motion & Physical Consistency]: There is an issue with the subject movement as the hand was having a distortion. Also with the subject object interaction as the  hand passed through to the object that subject was holding. Lastly with the cause-effect as the subject was not opening the packet of chips. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: The subject does not opens a blue and white chips packet. .\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/464.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman with short curly hair stands at a bathroom vanity under soft lighting. She wears a light blue T-shirt. Holding a tube of cream in her right hand, she squeezes a dollop onto her left palm. She then uses her right hand to gently rub the cream into the back of her left hand with smooth circular motions. The camera remains steady, showing her hands and the vanity countertop in the background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: video is blur [Motion & Physical Consistency]: hands are deformed, product is dispensed to her palms but not spread through her hands when she rubs it [Prompt Alignment]: she squeezes the tube of cream with her left hand ontp her right palm. she did not apply it in smooth circular motion\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/465.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A woman in her 30s with medium-length dark hair, wearing a light sweater and jeans, walks steadily through a brightly lit living room with simple furnishings while pulling a rolling suitcase behind her. She pauses near a coffee table, turns to face the camera, and waves with her right hand. The camera briefly focuses on the sides of the suitcase as light reflects off its surface, then returns to a steady shot of the woman and the living room..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: She was deformed. [Prompt Alignment]: She waves in her left hand.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/466.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged Latina woman with wavy dark hair stands in a room illuminated by soft, colored lights. She types on a keyboard set on a standing desk, her confident expression reflected in the faint glow of the surrounding panels. Slight traces of light seem to follow her fingers as she works efficiently..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The hand movement is not natural. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: The video is unrealistic.\n.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/467.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman stands confidently in a modern industrial workshop, shown from the waist up in a medium shot. She faces the camera directly, maintaining steady eye contact as she speaks clearly and authoritatively. She wears a dark blue work shirt with a mid-coverage neckline and covered shoulders, suitable for a manufacturing environment. The background is clean, with a slightly blurred metal workbench, a tool wall with hanging wrenches and screwdrivers, a red rolling toolbox, a tray of small assembly components, and orange safety gloves on the desk corner. The camera remains steady at her eye level, capturing her calm posture as she addresses the viewer without gestures..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/468.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Caucasian woman with short hair applies nourishing cream to her face in a cozy living room. The camera captures the smooth application as she gently massages it in, leaving her skin feeling moisturized and rejuvenated..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the subject movement as the hand of the subject was having stiffness. Also with the cause-effect as the cream was not evenly applies to the face. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/469.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Singaporean Chinese woman with a fair complexion stands in a softly lit bathroom, facing a mirror. She gently examines her reflection, focusing on the visible wrinkles around her eyes and cheeks. The camera captures both her face and the mirror image, highlighting the texture of her skin in natural lighting. She remains still, occasionally shifting her gaze slightly as she considers her appearance. The background shows a softly lit bathroom setting. The camera remains steady throughout the scene..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: There is no mirror and we cannot determine if she is facing in a mirror either.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/470.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black man with curly hair, wearing bright red athletic gear, sprints and leaps over a glowing blue high jump bar in a futuristic, cyberspace-inspired setting. His body arches in midair above the bar as neon lights pulse along the hovering track. The background features abstract digital patterns and vibrant colors. The camera follows his jump, capturing his movement against the illuminated environment..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: No [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: No - It looks like it was generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/471.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young East Asian woman sits at a vanity in a well-lit room. She holds out her forearm toward the camera, displaying seven different lipstick swatches arranged in a vertical line. The background shows a mirror and various makeup items on the vanity. The camera remains steady, focusing closely on her arm and the swatches..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Only six different lipstick swatches visible.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/472.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A close-up shot captures a hand placing an empty ceramic mug under a coffee machine spout; dark coffee begins to stream in, rising and forming crema at the top..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage [Motion & Physical Consistency]: The movement of the fingers appear distorted. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: The movement of the fingers appear distorted..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/473.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Latina woman in casual home clothes sprays thick white foam onto a greasy stovetop, watching it expand over the stained surface..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The way the foam being dispense is not accurate and the amount is not consistent. [Prompt Alignment]: The face is not visible.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/474.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A woman in a white lace dress sits on a couch in a modern cafe with large windows and outdoor seating visible in the background. She looks down at her large belly, gently adjusting her dress with both hands resting on her stomach. The camera remains steady, capturing her subtle movements and the contemporary setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Flickering issue on the dress. [Motion & Physical Consistency]: Deformed hand. [Prompt Alignment]: She is not gently adjusting her dress with both hands resting on her stomach.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/475.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Hispanic woman in a turquoise swimsuit stands on a high diving platform at a water park. She jumps off, tucking her body and twisting in mid-air while performing a somersault. Her arms and legs extend as she prepares to enter the water feet-first. The camera follows her movement from the side, capturing her form against the pool and surrounding park structures..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Yes [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: No - It looks like it was generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/476.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged Latino woman with a slim build and long curly hair sits in a cozy living room with wooden paneling and decorative lights. She wears a green shirt, layered necklaces, a headscarf, and large hoop earrings. She looks directly at the camera and speaks, her facial expressions showing subtle surprise as she reacts naturally to something unexpected. The camera remains steady, capturing her clear communication and composed demeanor in the relaxed setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/477.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A wide shot shows a tropical beach with golden sand and gentle sunlight. The ocean is filled with spaghetti noodles mixed with bolognese sauce, forming thick waves that roll toward the shore and break onto the sand. The camera remains steady, capturing the playful, surreal scene and the beach background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/478.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young South Asian woman kneels on a patterned rug in a dimly lit tent. She shuffles a deck of ivory-colored cards with embossed floral designs as wind whistles softly outside. A lantern beside her casts flickering light onto the richly embroidered fabric hanging behind her..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Good [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/479.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly South Asian man with wrinkled but steady hands crouches slightly on a sandy desert practice course as twilight bathes the surroundings in a golden glow. He adjusts his stance, his putter—a vintage copper-tone club—poised to tap a bright red golf ball toward a weathered hole flag surrounded by sparse patches of grass..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Good [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/480.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged East Asian woman with an elegant updo wears a flowing emerald green dress as she steps out of an enchanted forest bookstore. She carries intricately designed leather shopping bags embossed with glowing runes in her right hand. The air shimmers with soft golden light, and enchanted books hover on translucent shelves in the background, gradually fading as she moves forward. The last golden rays of sunlight illuminate her and the magical setting. The camera remains steady, capturing her graceful exit and the whimsical atmosphere..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: The subject was holding the shopping bag into her left hand.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/481.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: In a tranquil living room filled with soft, natural light, a young East Asian woman with short bobbed hair walks gracefully across the space, carrying a matte black rolling suitcase. Her steps are deliberate as she moves through the room, pausing by a plush sofa. She glances kindly at the camera, her hand lightly brushing against the suitcase handle. The camera zooms in for a close-up of the suitcase, revealing intricate designs etched into its trim, as the warm tones of the room create an inviting and relaxed atmosphere.\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage [Motion & Physical Consistency]: The movement of the hand appears stiff and not smooth. [Prompt Alignment]: She is not carrying it.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/482.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Black woman climbs a rope, her hands gripping firmly as she ascends the vertical challenge, sweat glistening on her skin..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Good. [Prompt Alignment]: The subject's face is not visible.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/483.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A white towel is pressed flat against the glass door of a spinning laundry machine. The machine rotates rapidly, causing the towel to move in a rhythmic circular motion. Soft, warm lighting fills the laundry room, and the camera remains steady, focusing on the towel and the spinning drum through the glass door..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The floating towel is unnatural. [Prompt Alignment]: The towel did not pressed flat against the glass door\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/484.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged man with short dark hair stands on a busy street corner, wearing a gray jacket and jeans. The camera smoothly zooms in over his right shoulder as he holds a smartphone in his right hand and looks down at the screen. The background shows urban buildings, passing cars, and pedestrians. The camera focuses on his hand and the phone as he interacts with the device, capturing the city atmosphere..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Flickering issue [Motion & Physical Consistency]: Deformed background. [Prompt Alignment]: He is not middle aged man and he did not holds a smartphone in his right hand and looks down at the screen.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/485.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Black man creates intricate latte art, swirling the milk into a delicate pattern on top of his morning coffee..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Cause-effect fails as milk appears and moves. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/486.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Middle Eastern woman holds a sweater near her face and gently sniffs it, visibly pleased by the fresh, just-washed scent..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The wordings are garbage symbols. [Motion & Physical Consistency]: Interaction with the object is not in accordance to physical law. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: I think it's generated by AI\n.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/487.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A hand places a minimalist black cup under a pod machine, and the camera stays fixed as hot coffee pours..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The coffee is supposed to pour out from the nozzle and the cause effect of the coffee is unnatural. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: I think it's generated by AI because of the unnatural movement..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/488.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The subject, a South Asian woman, carves a pumpkin, creating a smiling face for a festive fall celebration..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Good [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/489.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman with braided hair stands in a cozy, rustic bedroom with soft, warm lighting. She wears a loose sand-colored outfit. She spreads a bright turquoise comforter with geometric designs smoothly across a low wooden bed, using her hands to keep the edges neat. The background features simple, natural decor that contrasts with the vibrant bedding. The camera remains steady, capturing her actions and the inviting atmosphere of the room..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Her right hand or fingers are deformed. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/490.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly South Asian woman wearing a vibrant green sari stands beside a rustic stone tub outdoors, surrounded by baskets of freshly harvested crimson grapes under a clear blue sky. She carefully steps into the tub and balances herself as her bare feet press and squish the grapes. The camera remains steady, capturing her movements and the colorful grape-filled setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: No [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: No - It looks like it was generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/491.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The subject, a South Asian woman, picks a packet of pasta from the shelf, placing it into her shopping basket as she moves down the aisle..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The wordings in the product are just garbage, not English. [Motion & Physical Consistency]: Subject movement failure and Cause-effect failure. Stabilize to make it look normal. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: The subject is not South Asian woman..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/492.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A tan-skinned woman with short hair wears a tan blouse with small bubble sleeves at the shoulders and loose tan linen pants. She walks at a relaxed pace through the spacious interior of a luxury house, shown from behind as she moves past sleek modern furniture and tall windows filled with sunlight. The camera smoothly tracks her movement, highlighting her calm steps and the bright, elegant atmosphere of the home..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/493.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Black man grooms his dog, using a slicker brush to remove excess fur and ensure the dog’s coat is shiny and smooth..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Good [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/494.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young white woman with shoulder-length brown hair stands barefoot on wet sand beside tide-worn rocks at the beach during low tide. She wears rolled-up jeans and a light sweater. Holding a rusted hammer in her right hand, she firmly strikes an old, water-damaged phone placed on a flat stone. The phone’s screen cracks on impact, with small shards scattering across the rock. Gentle waves and wet sand are visible in the background. The camera remains steady, capturing the action and the beachside setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The deformed foot and the cause and effect of smashing seem unnatural. [Prompt Alignment]: The subject's face in the video is not visible, so we cannot identify whether she is a young white woman.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/495.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young man with light skin and short dark hair sits at a wooden desk in a cozy home office. He wears a casual blue sweater and uses his right hand to adjust numbers on a sleek silver laptop placed in front of him. The laptop screen displays a financial calculator interface with monthly payment figures and input fields. The camera starts with a close-up of the laptop, then slowly pans left to show more of the workspace and the man’s focused posture..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage symbols [Motion & Physical Consistency]: Good [Prompt Alignment]: The subject's face is not visible and he did not uses his right hand to adjust numbers on a sleek silver laptop placed in front of him.\nThe screen did not displays a financial calculator interface.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/496.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Southeast Asian man in his late twenties sits alone at a desk in a quiet, brightly lit office during overtime hours. He wears a light blue button-up shirt and black slacks. Holding his smartphone in his right hand, he scrolls through social media and pauses to watch a friend’s wedding story. He glances at the screen with a thoughtful expression, then looks up briefly before returning his gaze to the phone. The camera remains steady, capturing his subtle reactions and the quiet office setting with empty desks and paperwork around him..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Texture issues to the interface on the phone in the first part. [Motion & Physical Consistency]: There is deformity to the appearance of the person on the screen. Cause-effect fails as he didn't scroll the phone but it changed interface. [Prompt Alignment]: He is holding his smartphone in his left hand.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/497.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A muscular Japanese man in a white T-shirt and dark jeans stands in a bright kitchen. He casually pulls the refrigerator door wide open with both hands. Instantly, hundreds of baseballs tumble out, bouncing and rolling wildly across the kitchen floor. The man steps back slightly and stares at the scene in shock as baseballs scatter around his feet. The camera remains steady, capturing his reaction and the busy kitchen background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The texture of the design of the ball is flickering. [Motion & Physical Consistency]: Some balls change their form. Cause-effect fails as some ball suddenly appears. [Prompt Alignment]: He didn't pull the refrigerator door wide open with both hands.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/498.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Hispanic man in his 30s with short black hair wears a gray T-shirt and jeans as he vacuums the carpeted floor of his bedroom. He holds the vacuum’s hose in one hand and moves it steadily from one corner of the room to another, guiding the nozzle along the edges. The background shows a neatly made bed, a nightstand, and soft daylight coming through a window. The camera remains steady, capturing his cleaning motion and the tidy bedroom setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The vacuum is deformed. [Prompt Alignment]: He didn't hold the vacuum hose in one hand.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/499.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: Three young people dance together in a brightly lit indoor studio with exposed ceilings. The central dancer raises both arms overhead, forming a curved shape with the hands, then points both index fingers upward while stepping forward. The two dancers in the background mirror these arm and hand movements, stepping forward in sync with the central dancer. The camera remains steady, capturing the group’s coordinated movements..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Good [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/500.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Black man plays the cymbals with skill, the sound ringing through the concert hall as the band reaches its peak..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The subject's movements, interactions, and cause-and-effect are all unnatural. Make them look more natural. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: The subject's face in the video is not visible, so we cannot identify whether he is a Black man..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/501.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Hispanic man stands in a park, wearing comfortable gym clothing. He stretches his arms upward above his head, allowing his arms and upper body to move freely. The background shows grass and trees under daylight. The camera remains steady, capturing his stretching motion and relaxed posture..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Deformed hand is unnatural. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/502.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly Black woman with gray-streaked braids tied back neatly sits in the passenger seat of a vehicle. She wears a green knit cardigan and gestures and talks animatedly, waving her hand and then pointing firmly. Her gold-painted nails catch the light as she moves..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/503.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly South Asian man, dressed in flowing robes, strikes a poised stance on a serene desert dune as the twilight sky flushes with hues of orange and purple. One hand rests gently at his side while the other sweeps through the crown of his silver-streaked hair. His stance, with weight shifted to one leg, causes the billowing fabric around him to cascade and ripple in the soft evening breeze, lending an air of timeless elegance..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The overlay wordings are just garbage, not English. [Motion & Physical Consistency]: There is a distortion movement in the subject's hand. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: The subject is not South Asian..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/504.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A white teen in a beanie holds a mic in both hands and shares his thoughts during a group street interview, with friends chiming in over the city background noise..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: NULL [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/505.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The subject, a South Asian man, uses a squeegee to clean his windows, the glass gleaming as the water runs down the surface..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the cause-effect as the flow of water of the window was inconsistent and as it glides by the object it does not clean the surface of the window and the water flowing was to excessive and it was unrealistic being seen. . [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/506.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged Latina woman with dark hair stands on golden desert sand at dusk, wearing a flowing scarf. She confidently puts on a pair of sleek cat-eye sunglasses, then places her hands on her hips and looks ahead. The orange-pink light of the setting sun casts a warm glow over her and the quiet, expansive desert. The camera remains steady, capturing her poised stance and the serene landscape..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: good [Motion & Physical Consistency]: hand deformity during movement [Prompt Alignment]: Woman did notput on sunglasses but only adjusted it and no visual of hands on hips\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/507.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged man with short black hair, wearing a traditional Arabic robe and headscarf, stands in a well-lit living room. He faces the camera and speaks, maintaining a relaxed and confident posture. The background shows a sofa, a coffee table, and decorative cushions. The camera remains steady, capturing his upper body and the cozy indoor setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Texture issue around his beard and blurred lips [Motion & Physical Consistency]: Good [Prompt Alignment]: The man did not wear a headscarf\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/508.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: Three men in casual clothes lean forward on a couch, shouting and tapping buttons on controllers as the glow of the TV reflects on their faces..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The video has a garbage caption. [Motion & Physical Consistency]: There is an issue with the subject movement as the hand was having deformation also to his face. Also with the subject to object interaction as the hand was pressing the console but the console was not moving also the interaction of the subject towards to the object is unnatural. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: There is no TV .\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/509.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged Black man opens a spray bottle and places it into an olive-green pouch beside cleanser containers on a rustic bedroom table..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage text. [Motion & Physical Consistency]: His overall movements are unnatural. Interaction with the spray bottle when he opens it is unnatural. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: He didn't open the spray bottle..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/510.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A medium shot shows an American adult woman with long straight blond hair standing in a softly lit bedroom with light-colored walls and a wooden dressing table in the background. She wears a white nightgown and accessories on her left wrist. She holds an oral irrigator in one hand, raises it to her mouth, turns on the switch, and gently moves the device along her teeth to rinse her mouth. The camera remains steady, capturing her actions and the bedroom setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Interaction with the object as it is deform is not in accordance to physical law. [Prompt Alignment]: The accessories are on the right wrist.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/511.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A fair-skinned teenage boy with freckles and messy auburn hair stands in a bright, airy bedroom with pastel-colored walls. He wears a light blue hoodie and khaki pants. Holding a small, crystal-shaped microphone in his right hand, he gestures energetically toward a standing rack displaying casual clothing. The background features neatly arranged furniture and vibrant bedding, creating a cozy, youthful atmosphere. The camera remains steady, capturing his expressive gestures and the inviting room..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/512.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly man with silver hair and a wiry build stands in an overgrown forest ruin. Wearing a fitted T-shirt and athletic pants, he leaps with precision from one narrow stone wall to another, landing securely with both feet. Vines and moss cover the ancient stones, and golden morning light filters through the trees. The camera remains steady, capturing his agile movement and the lush, atmospheric setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Good [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/513.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young South Asian woman in athletic shorts and a T-shirt stands on a playground field. She steps forward and kicks a soccer ball with her right foot, sending it rolling across the playground toward a makeshift goal made of two cones. The background shows open space and playground equipment under bright daylight. The camera remains steady, capturing her movement and the ball’s path..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is a cause effect issue as the ball was moving on its own even though it was not kicked. [Prompt Alignment]: She can't be determined due to lack of view. The makeshift goal was not made of two cones.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/514.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Southeast Asian man in his 30s with short black hair sits at a kitchen table. He holds a golden-fried chicken wing in his right hand and dips it into a small container of thick honey garlic sauce. He lifts the wing, allowing some sauce to drip back into the container. The background shows a bright kitchen with simple decor. The camera remains steady, focusing on his hands and the dipping action..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Blurry foreground [Motion & Physical Consistency]: The cause-effect of the dripping sauce is unnatural. [Prompt Alignment]: It is not a chicken wing and he did not holds a golden-fried chicken in his right hand.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/515.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged man with short dark hair stands in a modern living room with light-colored walls and minimal decor. He holds a maraca in each hand and demonstrates how to play by shaking both maracas in a steady rhythm at chest level. The background shows a sofa and a floor lamp under soft indoor lighting. The camera remains steady, clearly capturing his demonstration and the contemporary setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Good [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/516.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: Inside a brightly lit modern kitchen, a young Black woman in a patterned apron carefully sets a steaming pumpkin pie next to an elegant spread of vibrant Thanksgiving dishes on a marble countertop. The table is adorned with bowls of creamy mashed potatoes, glazed carrots, and a golden turkey centerpiece. The aroma seems almost palpable, accompanied by the rich colors and textures of the food. A window reveals sunlit trees swaying outside..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Her hand is deformed, making it unnatural. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: I think it's generated by AI\n.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/517.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Hispanic woman in her 30s with shoulder-length dark hair sits at a kitchen table. She holds a fresh bread roll in her left hand and uses a butter knife in her right hand to spread soft butter evenly across the cut surface of the roll. The background shows a bright kitchen with a window and countertop. The camera remains steady, focusing on her hands as she spreads the butter..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/518.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A realistic handbag sits on a marble table—when sliced down the middle, layers of chocolate cake and frosting are revealed inside..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The movement appears non-sense. There seems no a little bit part of cake being cut or fall although the cake is being cut. There is no cut on the handle part although the edge of the knife touched it. [Prompt Alignment]: There is a finger visible. The cake was not sliced down the middle.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/519.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly man with glasses and a silvery beard prepares to bowl a cricket ball on a beach. He wears a crisp white shirt rolled up at the sleeves. The camera zooms in on the spinning ball as it leaves his hand, the waves crashing behind him in the early morning light..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Good [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/520.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A teenage boy with curly black hair and light brown skin struggles to push a boxy, white van up a steep dirt hill, surrounded by lush tropical trees swaying in the gentle breeze of the late afternoon..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: The subject is not  teenagers. The subject did not push a boxy, white van.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/521.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged white man with a rugged beard stands at the edge of an icy tundra, wearing a heavy winter jacket and gloves. He holds a frosty silver discus in his right hand. He exhales, creating a visible puff of vapor in the cold air, then throws the discus forcefully across the snowy landscape. The background shows a wide, frozen expanse under a pale sky. The camera remains steady, capturing his motion and the wintry setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: good [Motion & Physical Consistency]: good [Prompt Alignment]: good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/522.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The video showcases two Vaseline Pro Derma body lotion bottles being held by a pair of hands. The first bottle, labeled for dry skin, is adorned with blue accents and features ingredients such as 4D Hyaluronic Acid, Centella Asiatica, and Provitamin B5. The second bottle, designed for dull and uneven skin, has red accents and lists Niacinamide, Hexylresorcinol, and Vitamin C as its key ingredients. Both bottles are white with a pump dispenser and have the Vaseline logo prominently displayed.  The hands holding the bottles are positioned in a front view, with the camera capturing an eye-level, medium shot of the scene. The hands gently grip the bottles, showcasing them to the camera.  The background consists of lush green foliage, suggesting an outdoor setting. The leaves are large and vibrant, adding a natural and fresh ambiance to the scene. The text overlay reads \"MY FAVORITES Vaseline Pro Derma Body Lotions,\" emphasizing the focus on the products. Subsequent frames display the bottles lying horizontally on a flat surface, with sunlight casting shadows, highlighting the bottles' labels and design..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage symbols, not English. [Motion & Physical Consistency]: The subject's movements, interactions, and cause-and-effect are all unnatural. Make them look more natural. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: There are three bottles in the video, which are not stated in the prompt..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/523.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The subject, a South Asian man, pours his favorite cereal into a bowl, watching it fill up as he prepares for breakfast..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage [Motion & Physical Consistency]: The amount cereal increases although the only coming out from the packet is liquid. [Prompt Alignment]: He's not South Asian.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/524.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The hand gently holds the gold pendant, causing the chain to sway slightly and catch the light, showcasing the intricate details of the jewelry..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Good. [Prompt Alignment]: The chain didn't sway as the hand held it.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/525.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly South Asian man wearing a crisp white linen shirt and tan trousers walks out of a traditional open-air marketplace during golden hour. He carries several colorful woven shopping totes on his arms, with handcrafted jewelry visible through the bags. As he strides toward the entrance, the vibrant fabric stalls and market vendors recede in the background. Warm sunlight casts long shadows and highlights the details of his bags. The camera remains steady, capturing his movement and the lively market setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Good. [Prompt Alignment]: Only one arm carries the shopping totes.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/526.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young East Asian man with short black hair sits at a table covered with colorful candy and packaged sweets. He holds a skewer with alternating yellow, green, and orange round candies in his right hand. Next, he picks up a red and green strawberry-shaped lollipop with white polka dots, unwraps it, and reveals a pink and white lollipop inside, which he holds up in front of him. The table displays a pink cone-shaped candy package with cartoon graphics and a tray of large, brightly colored cylindrical candies. The camera remains steady, capturing the man’s actions and the vibrant assortment of treats..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: The subject's movement is stiff and looks robotic as the subject's movement is stiff, not smooth. When he ate the lollipop, it didn't get any smaller, but instead, it just changed shape. Stabilize to make it look normal. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/527.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged man with short dark hair sits at an office desk with his back to the camera. He wears a blue dress shirt and types on a computer keyboard. The office background shows a window with blinds and a shelf with books. The lighting is bright and natural. The camera remains steady behind him, capturing his upper body and workspace. then he looks down at his feet - camera angle changes with his vision.\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Good. [Prompt Alignment]: Good.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/528.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black man dots deep-toned liquid foundation across his face and buffs it in with a dense brush, revealing a smooth, even complexion as he turns to show both sides..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Cause-effect as the liquid foundation is not spread accurately. [Prompt Alignment]: No dense brush is visible. He didn't dot deep-toned liquid foundation across his face.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/529.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A medium shot shows a clean white ceramic plate at the center of a light wooden table. Fresh sushi pieces, including seaweed-wrapped rolls and nigiri with salmon and tuna, gently float downward from above and softly land on the plate in slow motion. Each piece settles, causing slight movement of rice grains. As the last sushi piece lands, the camera smoothly pans left to reveal the full lineup of sushi arranged neatly on the plate. The background is softly lit, with the wooden tabletop visible..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: textural issues [Motion & Physical Consistency]: the last roll that dropped had 2 seaweed then when it landed, it became to one piece of seaweed [Prompt Alignment]: The sushi pieces did not gently float downward, instead it dropped chaotically, no slight movement of rice grains\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/530.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young White woman with an athletic build sits confidently in a modern home studio, softly illuminated by professional lighting. She wears a modest game character costume with sleek armor details and vibrant colors. Her expression is calm and composed as she maintains steady eye contact with the camera, speaking directly to the viewer in a clear, natural tone. Her movements are minimal and deliberate, emphasizing a focused conversation. The background features gaming memorabilia and streaming equipment, creating a warm and inviting esports atmosphere. The camera remains fixed at eye level, capturing her upper body and the studio setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the subject interaction as the hand of the subject was seen having a deformation. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/531.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: In a cozy home kitchen, a Middle Eastern man in a white shirt scoops cooked lamb from a large casserole dish. The camera focuses on the tender, shredded meat as it is placed on a plate, ready for the dinner table..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is a cause effect issue as the amount of lamb being put in the plate does not match the amount he scooped. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/532.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A white teenage girl creates a bubbly foam in her hands and rubs it onto her face, the suds covering her nose and cheeks as she smiles into the mirror..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The hand movement and the cause and effect of the bubbles are unnatural. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: No. Object Existence: No. Subject-Object Interaction: The subject in the video is not a white teenage girl and it is not shown in the video that she creates a bubbly foam in her hands..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/533.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young East Asian woman with straight black hair stands on a simple bridge at night, illuminated by soft ambient light. She wears a flowing dress that drapes naturally around her waist. Her posture curves gently, accentuating her silhouette, with her arms relaxed at her sides. She looks calmly at the camera and lifts one hand to blow a kiss into the air. The camera remains steady, capturing her graceful movement and the quiet nighttime setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Deformed hand [Prompt Alignment]: She did not lifts one hand to blow a kiss into the air.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/534.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A wide sweeping camera shot moves from a balcony overlooking a lush valley toward a large distant volcano. Fog drifts gently below, partially obscuring the greenery on the hillsides. The sky is overcast, casting soft, diffused light across the landscape. The balcony features a simple railing in the foreground, and the camera glides smoothly, capturing the dramatic scenery and misty atmosphere..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The texture of the tree is flickering. [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/535.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A female hair stylist with shoulder-length dark hair, wearing a black salon apron, stands behind a seated woman with straight, shoulder-length brown hair. The stylist uses a round brush in her left hand and a blow dryer in her right hand to brush and blow-dry sections of the client’s hair. The background shows a modern salon with large mirrors and bright lighting. The camera remains steady, capturing the stylist’s hands as she works on the client’s hair..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: The stylist uses a round brush in her right hand and a blow dryer in her left hand to brush and blow-dry sections of the client’s hair.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/536.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Malay woman wearing a hijab sits at a cozy cafe table. She holds a strawberry matcha drink in one hand and faces the camera, smiling softly. The video is filmed from her phone camera, showing a close-up of her face and the drink. The background is softly blurred, with warm indoor lighting and cafe decor. The camera remains steady, capturing her relaxed posture and subtle movements..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/537.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The camera pans across a line of people outside a theater, passing a Middle Eastern man in a long coat, a white woman clutching a program, a Black girl with braids and a bright scarf, and a Southeast Asian boy holding popcorn..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage [Motion & Physical Consistency]: The movement appears deformed and distorted. The way the black girl interact or hold the scarf is not in a natural way. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: There are other subject that wasn't mentioned. The movement appears deformed and distorted. The way the black girl interact or hold the scarf is not in a natural way..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/538.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Black man with short hair stands in a softly lit bathroom. He holds a serum dropper above his nose and squeezes a drop of serum directly onto the bridge, where textured pores are visible. The serum spreads naturally across his skin as he tilts his head slightly. The camera remains steady, focusing on his face and the gentle lighting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Deformed hand and the cause-effect of the dripping serum is unnatural. [Prompt Alignment]: He did not squeezes a drop of serum directly onto the bridge, where textured pores are visible.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/539.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The hand gently holds the Geek Bar device, moving smoothly and slightly to display its surface and details, with a subtle zoom-in to highlight the product’s design against the colorful background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The tip of the finger in the left hand glitches. [Prompt Alignment]: It isn't a Geek bar device.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/540.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A white teenage girl with long, damp hair lathers shampoo at the crown of her head, her arms raised as suds spread through her strands in a soft backlit shower..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Good. [Prompt Alignment]: Good.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/541.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Southeast Asian woman with long black hair and a bright smile waves both hands enthusiastically toward the camera while sitting in a softly lit bedroom..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage text. [Motion & Physical Consistency]: Her overall movements when waving are unnatural. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Unnatural movements..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/542.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A close-up captures a white woman’s lips slightly parted as she swipes on a coral glossy lipstick, creating a smooth, glass-like finish..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Her movement of applying lipstick is inconsistent. Her fingers holding lipstick are unnatural. The effects of lipstick applied to her lips are unnatural. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/543.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: Two adult men dressed in casual urban clothing stand on a set of concrete steps in an urban area with tall buildings in the background. Both men are wearing helmets and gloves. The man on the left jumps off the steps, performing a trick in mid-air, while the man on the right skates along the steps. A yellow taxi and other vehicles are visible on the street below. The camera follows the two men as they move through the city, capturing their actions and the urban setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: There is a flickering issue with the buildings. [Motion & Physical Consistency]: The deformed stairs and the cause and effect of the skates seem unnatural. [Prompt Alignment]: The action in the prompt is not followed in the video.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/544.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young South Asian man with curly hair sits in the passenger seat of a modern car with black leather seats and a panoramic sunroof. He wears a navy blue hoodie. Sunlight streams through the glass, highlighting his hair. He gestures animatedly while speaking, moving his hands in wide arcs. Through the partially opened car window and door, a serene beach and sparkling blue ocean are visible beyond a fence. The camera remains steady, capturing his expressive gestures and the bright, coastal background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: There is deformity in the hand. [Prompt Alignment]: He's not South Asian. The car window and door are not partially open.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/545.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A white teenage boy stands barefoot in front of a full-length mirror, holding a t-shirt and jeans against his chest with a sleepy expression as sunlight streams in..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The movement of the boy is not natural as his hands move stiffly. There is also an object-object interaction issue, as the t-shirt and jeans appear merged. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: The movement of the boy is not natural as his hands move stiffly. There is also an object-object interaction issue, as the t-shirt and jeans appear merged. I believe the video is AI-generated because of this discrepancy..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/546.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman waxes her legs, the wax applied smoothly as she prepares for a flawless finish..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Cause-effect failure as the wax didn't spread accurately. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/547.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Southeast Asian woman stands by a window in a sunlit living room. She holds a can of foaming cleanser in her right hand and aims it at the dusty window track. She presses the nozzle, spraying a stream of foam directly into the narrow, dirty groove until the foam quickly fills the space. The background shows light curtains and part of a sofa. The camera remains steady, focusing on her hand and the window track..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is a cause effect issue as the foaming cleanser appeared inconsistently and excessively. [Prompt Alignment]: The subject can't be determined due to lack of view.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/548.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young white woman with shoulder-length brown hair stands barefoot on wet sand beside tide-worn rocks at the beach during low tide. She wears rolled-up jeans and a light sweater. Holding a rusted hammer in her right hand, she firmly strikes an old, water-damaged phone placed on a flat stone. The phone’s screen cracks on impact, with small shards scattering across the rock. Gentle waves and wet sand are visible in the background. The camera remains steady, capturing the action and the beachside setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Texture on the rock. [Motion & Physical Consistency]: The screen cracks but then it back to normal again. [Prompt Alignment]: Cannot determined if woman is young and white with shoulder-length brown hair.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/549.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Black man mixes ingredients for a marinade in a bowl, preparing to coat his chicken with the flavorful mixture..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The spoon passes through the chicken. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/550.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A stylish Black woman in a long beige trench coat over a black top and jeans stands confidently on a cobblestone street in SoHo. She shifts her weight slightly and lets her trench coat flow in the breeze. The background features cast-iron buildings with large windows. The camera remains steady, capturing her poised stance and the urban setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/551.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A woman with long, straight hair is standing in a room with a light-colored wall and some furniture in the background. She quickly goes from young to old during the video. She is wearing a white lace top over a white t-shirt, denim shorts, and multiple gold necklaces and bracelets. She is also wearing rings on her fingers and a gold watch on her left wrist. The woman is gesturing with her hands while talking, moving her hands up and down and occasionally bringing them close to her chest..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: It is not shown in the video that she quickly goes from young to old during the video.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/552.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young adult with light skin and neatly trimmed nails sits in a dental chair, wearing a blue dental bib. The subject smiles widely, exposing their teeth. A dental professional wearing blue gloves holds a small dental mirror close to the subject’s mouth, positioning it near the lower teeth. The subject’s tongue is partially visible at the bottom of the frame. The background shows a dental office with bright lighting and medical equipment. The camera remains steady, focusing on the subject’s face and the dental mirror..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The deformed dental mirror is unnatural. [Prompt Alignment]: It is not a dental mirror. The subject's face in the video is not visible, so we cannot identify whether she is a young adult. The object is not  position near the lower teeth.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: No. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/553.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Caucasian woman with freckles and red hair stands beside her bedside table in a softly lit bedroom. She wears a light-colored long-sleeve shirt. She raises her right index finger and points toward an empty space above the table. Cozy blankets are visible on the bed, creating an inviting, calm atmosphere. The camera remains steady, capturing her gesture and the warm bedroom setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Her fingers are distorted. [Prompt Alignment]: No table is visible. She raised her left index finger.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/554.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young woman sits at a restaurant table with a plate of food in front of her. Another person’s hand appears from the right, holding a piece of food and bringing it toward her mouth. The woman remains seated and still, while the hand continues to bring the food closer to her. The background shows a busy restaurant setting. The camera remains steady, capturing the interaction at the table..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage symbols [Motion & Physical Consistency]: The bite size of food doesn't decrease in her mouth. [Prompt Alignment]: The food was already in her mouth. The hand didn't bring it towards her mouth.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/555.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The subject, a South Asian woman, applies a bright eyeshadow color with a makeup brush, smoothing it onto her eyelid for a vibrant, dramatic effect..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the cause-effect as no visible changes was seen upon applying the object to the upper of the eyelid. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/556.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged South Asian man stands at a slate-grey marble countertop in a kitchen. His hands are lightly dusted with flour. He uses a pastel pink spatula in his right hand to slowly and steadily spread a creamy white substance across the countertop. The background is softly lit by dim, golden evening light, which makes the cream glisten. The camera films from above, focusing on the spatula’s smooth motion and the texture of the cream. The scene is minimalistic, with the countertop and the man’s hands as the main focus..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The hand movement and the cause-effect of the cream is unnatural. [Prompt Alignment]: The subject's face is not visible.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/557.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Caucasian man runs energetically through a wheat field, his arms swinging as his legs stretch with each stride. The camera captures his determination as the golden wheat parts around him, the field stretching out endlessly in front..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Subjects movement is unnatural. Stabilize to make it look normal. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: I think it's generated by AI because of the unnatural movement..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/558.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An energetic young Latino boy in a sky-blue shirt leaps off a wooden platform, his arms outstretched to catch the swinging trapeze in mid-air. The camera zooms in on his gleeful face, framed by the bright carnival lights and colorful pennant flags behind him..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Good [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/559.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young White man with curly hair stands on a bustling city rooftop at sunset. He wears a crisp white shirt and ties a matte black necktie around his collar, using both hands to form the knot. The fiery orange skyline glows behind him as the wind gently tousles his hair. The camera remains steady, capturing his focused expression and confident hand movements..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Hand movement appears robotic and the tie and his hands deforms while he's tying. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/560.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly South Asian man wearing a traditional embroidered shawl stands on a bustling city rooftop at twilight, illuminated by flickering neon signs. He steps back with a calm posture, letting his arms rest naturally at his sides. He looks directly at the camera, then slowly leans forward and lifts one hand to blow a kiss. The background shows city lights and distant buildings. The camera remains steady, capturing his gentle movement and the urban evening atmosphere..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: He did not step back.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/561.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: On a windswept sandy beach, a middle-aged Latina woman with weathered yet gentle features spins earthy-toned yarn on a portable spinning wheel. Her hands move fluidly, creating long threads while ocean waves crash softly in the background, and the setting sun casts golden light on her craft..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Interaction with the object is not in accordance to physical law. [Prompt Alignment]: Good.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/562.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A South Asian man stands in a well-lit bathroom, holding an electric trimmer in his right hand. He carefully moves the trimmer along the contours of his beard, maintaining a steady hand as he shapes the facial hair. The background shows a mirror and bathroom tiles. The camera remains steady, focusing on his grooming technique..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Yes [Motion & Physical Consistency]: Cinematic Style - (Select Cinematic if the video looks like a professionally shot film scene) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No - The video does contain nudity or harmful content. Cause-Effect: No - It looks like it was generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/563.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman with braided hair sits at a wooden desk in a well-lit room. She holds a pencil in her left hand and uses a manual sharpener in her right hand to shave the tip off the pencil. The wood shavings fall onto the desk as she prepares for a drawing session. The camera remains steady, focusing on her hands and the desk surface..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Interaction with the pencil and the sharpener is not in accordance to physical law. Cause-effect failure as the size of the pencil is still the same. [Prompt Alignment]: She didn't hold a pencil in her left hand and didn't use a manual sharpener in her right hand.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/564.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly South Asian man with a neatly trimmed white beard stands under the dappled light of a large oak tree in a peaceful park. He wears a light button-up shirt and dark trousers. He uses one hand to adjust his sleek, matte-black sunglasses, then lowers his hand and stands with a relaxed, confident posture. The background shows lush green grass and leafy trees, with soft sunlight filtering through the branches. The camera remains steady, capturing the tranquil park setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/565.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly man with gray hair and a neatly trimmed beard stands in a well-lit luxury retirement home. He holds a bouquet of flowers in both hands and approaches an elderly woman seated at a table with her friends. The woman, wearing a patterned dress, smiles and reaches out to accept the flowers. The background shows other tables and chairs in the cozy room. The camera remains steady, capturing the interaction and the festive atmosphere..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/566.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly South Asian man with a wispy gray beard, dressed in crisp linen pajamas, refreshes his bedroom at dawn. The room is bathed in soft, golden light filtering through grand arched windows. He spreads a luxurious silk comforter adorned with intricate patterns across a wooden canopy bed. His hands glide across the surface, smoothing wrinkles with intent precision. A gentle breeze rustles a nearby vase of fresh jasmine, adding an aromatic layer to the serene, elegant atmosphere of the classic bedroom setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: There is garbage text visible in the video. [Motion & Physical Consistency]: The man's movement appears unnatural, as his hands move stiffly. There is also a subject-object interaction issue, as the comforter changes shape and deforms unnaturally. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: No. Object Existence: No. Subject-Object Interaction: The man is not South Asian. The lighting is not dawn. The man's movement appears unnatural, as his hands move stiffly. There is also a subject-object interaction issue, as the comforter changes shape and deforms unnaturally. I believe the video is AI-generated because of this discrepancy. .\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/567.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged man with light brown skin and salt-and-pepper hair stands on a sandy beach at sunset, wearing a crisp white shirt. Gentle waves lap at the shore behind him, and the sky glows with golden light. He arches his torso slightly, letting his arms rest loosely at his sides in a relaxed, open stance. He looks at the camera with a kind expression, then lifts one hand to blow a kiss. The camera remains steady, capturing the peaceful beach and warm sunset atmosphere..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: He did not blows a kiss.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/568.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young woman with deep brown skin and braided hair kneels on a grassy field in a vibrant park at sunset. She holds a small jar of translucent green cream in one hand and uses her fingers to spread the cream onto a scrape on her leg. Her playful soccer ball rests nearby. The background features colorful trees and warm evening light. The camera remains steady, capturing her actions and the lively outdoor setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The effects of the cream when massaging are unrealistic. [Prompt Alignment]: She isn't kneeling.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/569.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman with curly hair wears a pair of sleek, reflective sunglasses as she leans confidently against a weathered brick wall in a sunlit urban alley. The vibrant graffiti behind her adds a burst of color, while her cool demeanor radiates through her subtle yet commanding pose, the sunlight glinting off the lenses of her shades..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the subject movement as the hand of the subject was having deformation. Also with the subject object interaction as the hand of the subject went through to the object she was holding and interacting. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: There is an issue with the subject movement as the hand of the subject was having deformation. Also with the subject object interaction as the hand of the subject went through to the object she was holding and interacting. .\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/570.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young white woman stands beside a small brown and white horse tied to a brick wall in an outdoor area. She bends over and uses a blue grooming brush on the horse’s brown side near its front leg. After brushing, she straightens up, places her left hand on the horse’s neck, and holds the brush in her right hand. The horse stands calmly, secured to the wall with a blue halter and lead rope. The camera remains steady, capturing the grooming action and the outdoor setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The hand movement, the cause-effect of the fur and the sudden changes of the object are all unnatural. [Prompt Alignment]: She did not places her left hand on the horse’s neck, and holds the brush in her right hand.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/571.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: Four women pose individually in front of the camera, each highlighting their outfit. First, a Southeast Asian woman in a belted jumpsuit steps forward and poses. Next, a Black woman in a statement blazer and trousers takes her turn. Then, a white woman in a long knit dress poses. Finally, a Latina woman in a monochrome co-ord set poses. The background is a simple, well-lit studio. The camera remains steady, capturing each woman as she poses in sequence..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The deformed face seem unnatural. [Prompt Alignment]: The video is a split screen which is not mentioned in the prompt. There is no Black woman\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: No. Object Existence: No. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/572.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Southeast Asian woman sits in the passenger seat, holding a tray of colorful sushi in her hands, showcasing each piece to the camera with a cheerful smile..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: There is a flickering issue with the trees. [Motion & Physical Consistency]: The hand movement and the interaction with the lid or wrapper are unnatural. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: In the video, the subject is not sitting in the passenger seat..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/573.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: Under the cool fluorescence of a modern clinic, a young Black woman, dressed in crisp, casual attire, tenderly holds her elbow, exposing dry patches on otherwise smooth skin. The textured appearance of the elbow comes into sharp focus, framed by clean white surfaces and medical charts in soft blur behind her. The overlay 'if you have this' appears in sleek black font, complementing the clinical atmosphere. Her hand's graceful motion draws attention to the skin’s dryness and delicacy..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The wordings are garbage symbols. [Motion & Physical Consistency]: Good. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: The text does not match. She didn't hold her elbow..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/574.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A wide shot shows a realistic-looking empanada riding a giant wave on a blue ocean near the shore of a Costa Rican city. Tall modern buildings and green palm trees are visible in the background. The camera tracks sideways, keeping the empanada and surfboard centered as the wave carries them closer to the sandy beach..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Flickering issue [Motion & Physical Consistency]: Deformed body [Prompt Alignment]: It does not look like an empanada\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/575.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Black teenage girl with braided hair stands in front of a mirror in a softly lit bedroom. She holds a pastel-colored cat-ear headband in both hands and gently places it on her head, adjusting it slightly. She smiles at her reflection as she finishes. The background shows a neatly made bed and a small dresser. The camera remains steady, capturing her actions and the cozy room..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Her finger passed through and the reflection in the mirror are seem unnatural. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/576.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Black woman climbs a rope, her hands gripping firmly as she ascends the vertical challenge, sweat glistening on her skin..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: NULL [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/577.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A college-aged woman with long dark hair stands beside a messy bed in a small dorm room. She holds a closed laptop in one hand and drops it onto the bed, where it lands among scattered notebooks and tangled sheets. The laptop lid bounces slightly upon impact. The background shows a cluttered nightstand and posters on the wall. The camera remains steady, capturing the action and the disorganized setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is a subject interaction issue as the hand of the subject is all right. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/578.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A preteen girl with braided hair and glasses, wearing a colorful sweater, smiles mischievously as she rolls large blue and red dice onto the Monopoly board. She sits at a round table, surrounded by her siblings, in a brightly lit family dining room..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: NULL [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/579.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young man in a tailored suit stands in a spacious, modern luxury home with floor-to-ceiling windows and sleek marble flooring. He faces the camera, holding his phone in his right hand. He begins to walk forward along a hallway lined with minimalist furniture and large abstract artwork, moving steadily past the pieces. Sunlight streams through the windows, illuminating the open living area behind him. The camera follows him at a slow pace, capturing the elegant interior and architectural details..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The face of the man deforms. [Prompt Alignment]: The phone is in his left hand and not his right hand.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/580.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The subject, a South Asian woman, climbs a rope outdoors, the morning sunlight casting a warm glow on her as she scales the rope with ease..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: NULL [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/581.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Middle Eastern woman with a ponytail walks through a park filled with autumn leaves in a forest green colored hoodie. The camera follows her from behind, showcasing the relaxed fit and cozy feel of the hoodie as she enjoys the crisp fall air..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The overall walking movement of the subject is unnatural. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: The camera follows her from her side not behind her..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/582.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly Black man with a graying beard sits on a wooden bench in a quiet park filled with autumn leaves. He carefully opens a matte black drone box, lifts the drone out, and examines its intricate details as sunlight reflects off its polished surface. The camera remains steady, capturing his focused expression and the peaceful park background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: 0004 item in box appears to be pixelated and blurry [Motion & Physical Consistency]: good [Prompt Alignment]: subject did not carefully opens a matte black drone box as it is already opened. Also no drone can be seen in the video.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/583.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged woman with auburn hair and freckled skin crouches beside a flickering campfire on a grassy meadow at sunset, carefully seasoning a pan of sizzling fish, the vibrant greenery blending seamlessly with the deep orange horizon..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The movement and interaction of the subject with the object appear unnatural, lacking fluidity and disrupting realism. Additionally, there is a cause-effect issue, as the seasoning were inconsistent. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: The movement and interaction of the subject with the object appear unnatural, lacking fluidity and disrupting realism. Additionally, there is a cause-effect issue, as the seasoning were inconsistent.  I believe the video is AI-generated because of this discrepancy. .\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/584.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Middle Eastern woman wearing a red cap and matching uniform stands in a brightly lit kitchen. She smiles as she uses both hands to lift open the lid of a fresh pizza box, revealing a cheesy margherita pizza inside. The background shows warm kitchen lighting and countertops. The camera remains steady, capturing her action and the pizza..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: The subject is not clearly visible to identify as  Middle Eastern woman.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/585.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A close-up macro shot shows swirled chocolate and vanilla ice cream with glossy fudge ripples running through the surface. Tiny air pockets are visible in the creamy texture. The camera remains steady, focusing on the detailed patterns and smooth, shiny appearance of the ice cream. The background is softly blurred to highlight the dessert..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/586.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly South Asian man with a wispy gray beard, wearing crisp linen pajamas, stands in a classic bedroom at dawn. Soft golden light streams through grand arched windows, illuminating a wooden canopy bed. He spreads a silk comforter with intricate patterns across the bed, using both hands to smooth out wrinkles with careful, precise movements. A vase of fresh jasmine sits nearby, and a gentle breeze stirs the flowers. The camera remains steady, capturing the serene and elegant atmosphere of the room..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Deformed fingers. [Prompt Alignment]: It does not look like a jasmine flower.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/587.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: POV from inside a shopping bag: a Black woman’s hand reaches in and grabs a boxed skincare product, her nails painted a glossy nude..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: NULL [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/588.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A white teenage boy sits on a couch in a living room, holding a bowl of mac and cheese in his left hand and a fork in his right. He uses the fork to scoop up mac and cheese and brings it to his mouth, keeping his eyes focused on the TV screen in front of him. The background includes a coffee table and soft lighting. The camera remains steady, capturing his relaxed posture and eating motion..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The hand movement of the subject is unnatural. Sudden movement of the mac is unnatural. [Prompt Alignment]: There is no TV screen visible and he is holding the mac cheese in his right hand.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/589.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A teenage Asian woman with a ponytail stands on a grassy field in bright late afternoon sunlight. She grips a black field hockey stick with both hands and swings it to strike a bright orange ball, sending the ball rolling quickly across the grass. The camera remains steady, capturing her focused movement and the sunlit field in the background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: No [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: No - It looks like it was generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/590.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young White woman with an athletic build sits confidently in a modern home studio, softly illuminated by professional lighting. She wears a modest game character costume with sleek armor details and vibrant colors. Her expression is calm and composed as she maintains steady eye contact with the camera, speaking directly to the viewer in a clear, natural tone. Her movements are minimal and deliberate, emphasizing a focused conversation. The background features gaming memorabilia and streaming equipment, creating a warm and inviting esports atmosphere. The camera remains fixed at eye level, capturing her upper body and the studio setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Texture issue on the eyebrows and lip / Garbage symbol [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/591.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: Set in a serene forest clearing at sunrise, a middle-aged woman with long, flowing brown hair and fair skin sits on a tree stump. She holds a stack of earthy-toned shoe boxes with green leaf illustrations, her expression glowing with delight. Sunbeams cascade through the trees, and the soft sounds of birds make her movements to explore the boxes feel magical and connected to nature..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The movement of the woman's hands while holding the boxes is not natural, as they appear deformed. There is also an object-object interaction issue, as the box she is holding merges with the box below it. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: The movement of the woman's hands while holding the boxes is not natural, as they appear deformed. There is also an object-object interaction issue, as the box she is holding merges with the box below it. I believe the video is AI-generated because of this discrepancy. .\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/592.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The subject, a South Asian woman, inflates a balloon with a deep breath, the round shape forming quickly as she ties the end..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Subject interaction fails as other balloons appear. Cause and effect fail as balloons pump air in its own very quickly. Stabilize to make it look normal. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/593.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A group of food items—pizzas, burgers, fries, and sushi—are arranged on a road marked with a double yellow line. Each food item has legs and arms, and they walk together along the road under a cloudy sky. The background features a row of trees lining the roadside. The camera remains steady, capturing the movement of the food items as they progress forward..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Deformed legs and the cause-effect of the moving food items are unnatural. [Prompt Alignment]: It looks like they are not walking.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/594.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A medium shot shows a pair of light-skinned adult hands holding an iPhone against a plain white tabletop. The fingers wrap securely around the phone, with the thumbs gently tapping and swiping across the screen. The wrists rest on the surface as the device is slightly repositioned for a better angle. The camera slowly pans to the right, focusing on the hands and the phone..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: texture issues on back of hand [Motion & Physical Consistency]: left hand deformed when interacting with phone [Prompt Alignment]: the index finger is the one that is intereacting with the phone, not the thumbs, wrists did not reposition.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/595.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A digital animation illustrates the origin of the universe. The scene opens with complete darkness, representing absolute nothingness. Suddenly, a burst of radiant energy appears at the center, rapidly expanding outward. The expansion is silent, showing space itself stretching at incredible speed. Swirling patterns of energy transform into tiny glowing particles—protons, neutrons, and electrons—floating and clustering together. The background gradually shifts from black to deep blue, hinting at the birth of cosmic structure. The camera slowly zooms out, revealing the growing expanse filled with fundamental particles..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: It is not swirling patterns of energy as it transforms.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/596.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young adult with short dark hair sits at a desk in a modern, well-lit home office. They face the camera and speak, holding a smartphone in one hand to display the XM platform. The background includes shelves with books and a computer monitor. The camera remains steady, capturing the subject’s upper body and the office setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage Symbols. [Motion & Physical Consistency]: Good. [Prompt Alignment]: No computer monitor and it is not an XM Platform.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/597.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Middle Eastern woman with long dark hair stands at a vanity in a softly lit bedroom. She uses a dropper to apply golden-toned foundation drops across her face, then blends the foundation evenly with quick, confident strokes using her fingertips. She leans closer to the mirror, turning her head slightly to show her smooth, even skin. The camera remains steady, capturing her actions and the cozy bedroom background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the cause-effect as the golden-toned foundation was not evenly applies to the face and does not scattered along the way of the face even if it was glided making it appear as inaccurate upon application. [Prompt Alignment]: The subject does not using fingertips to blends the foundation.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/598.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged Latina woman wearing a sunhat stands outdoors, surrounded by vibrant green forests. She holds a ceramic egg separator shaped like a small bird above two plates. Using both hands, she cracks an egg and pours its contents into the separator, allowing the egg white to flow onto one plate while the yolk remains in the separator and is then transferred to the second plate. The camera stays steady, capturing her actions and the lush forest background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: The egg yolk multiplies. Stabilize to make it look normal. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/599.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The person subtly shifts his posture while wearing the white patterned headscarf, allowing the fabric’s intricate texture and design to catch the light..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The wordings are garbage symbols. [Motion & Physical Consistency]: Good. [Prompt Alignment]: Good.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/600.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young girl with a pink frock fills a glass with water from a bright blue pitcher in her cozy kitchen. The camera captures her hand as she lifts the pitcher, pouring water into the glass, her face lit up with excitement..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: NULL [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/601.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Caucasian woman with shoulder-length brown hair lies in a hammock outdoors, her eyes closed as the hammock gently sways. The background shows green grass and trees under soft daylight. The camera remains steady, capturing her relaxed posture and the subtle motion of the hammock..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Yes [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: No - It looks like it was generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/602.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A White adult man stands on a grassy area near a pond, holding a remote control device with both hands. He smiles and looks down at the device, gently adjusting it with his fingers. The background features a calm pond surrounded by lush greenery and a well-manicured lawn. The camera remains steady, capturing the man’s actions and the peaceful outdoor setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Texture issue on the beard / Blurred hand [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/603.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged White man with a slim build and glasses sits at his office desk in a modern workspace. He wears a checkered long-sleeve shirt. Large windows behind him show buildings in an urban setting. He looks directly at the camera, maintaining steady eye contact and a calm, professional expression. The room is quiet and empty, with no other people present. The camera remains fixed at eye level, capturing his upper body and the office background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Texture shirt. [Motion & Physical Consistency]: Good [Prompt Alignment]: The man is medium build, not slim build.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/604.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly Black woman in a flowy dress sits on a park bench surrounded by lush greenery. She holds a large yellow calculator with pronounced keys in her hands and presses the buttons as she tallies market purchases, smiling gently. The background shows trees and bushes in soft morning light. The camera remains steady, capturing her actions and the peaceful outdoor setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage symbols on the buttons. [Motion & Physical Consistency]: Her fingers are glitching when pressing the buttons. [Prompt Alignment]: She isn't tallying market purchases since the caculator isn't turned on either.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/605.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Black man with short hair stands in a softly lit bathroom. He holds a small dropper bottle in one hand and uses the dropper to apply a drop of serum directly onto the bridge of his nose, where textured pores are visible. The serum spreads across his skin as he gently moves the dropper tip. The camera remains steady, focusing on his face and the application process..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the cause-effect as the serum was seen dropping onto the face even though the dropper contained no visible serum, and it did not dispense properly. Additionally, the serum was not applied evenly, with drops appearing inconsistently across the subject’s face, making the interaction appear unrealistic. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/606.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Southeast Asian man runs his fingers through his wet hair while rinsing shampoo out, the camera capturing the back view as water flows down his neck..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Movement is stiff and not smooth.  Water flows in his back suddenly disappears. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/607.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black man holds a takeout container of crispy fried chicken and fries on his lap in the driver's seat, sunlight hitting the golden crust as he lifts the lid..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the subject movement as the hand of the subject was having a stiffness. Lastly with the subject object interaction as the food went through also the lid of the object was seen smaller and it was not accurate as the container was big but the lid was appearing to be small. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: The subject was not position at the driver seat as no steering wheel was seen to be recognized as it was on the driver seat. .\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/608.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged Eastern European man sits in a reclining dental chair in a brightly lit dental office. A young female dentist in a white coat stands beside him, adjusting her gloves and preparing dental tools on a tray. The dentist leans over the man’s open mouth, using a small mirror and probe to examine his teeth. The camera remains steady, capturing the dental examination from a medium angle as the dentist carefully checks the man’s teeth. The background shows dental equipment and bright overhead lighting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: blurry hands 0001 [Motion & Physical Consistency]: deformed hands and subject movement when using tool to interact with client teeth looks weird. [Prompt Alignment]: good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/609.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Hispanic man stands in a modern kitchen with white cabinets and a marble countertop. He opens the stove door with his right hand, then glances at the camera with a slight smile. The background shows bright, natural lighting and kitchen appliances. The camera remains steady, capturing his actions and the kitchen setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Deformed hand [Prompt Alignment]: There is no kitchen appliances visible and he is not Hispanic man.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: No. Object Existence: No. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/610.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Black Brazilian man with deep brown skin stands in front of a bathroom mirror. He wears a gray T-shirt. He dispenses serum onto his fingertips and applies it to his forehead, then uses gentle circular motions to smooth the serum into his skin. The background shows a tiled wall and bathroom lighting. The camera remains steady, focusing on his face and hands as he completes the skincare routine..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Texture issue on the face skin [Motion & Physical Consistency]: Good [Prompt Alignment]: There is no mirror or serum. The man did not dispense serum onto his fingertips and applied it to his forehead\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/611.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Arabic woman with long dark hair sits in the driver’s seat of a sleek, modern car. She wears a modest light grey blouse and matching trousers. The brown leather headrest and softly illuminated dashboard are partially visible in the background. She looks directly at the camera, speaking with a calm, steady tone and subtle expressions that convey dissatisfaction. The camera remains fixed at eye level, capturing her upper body and the car’s interior details..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: texture issue of hair [Motion & Physical Consistency]: good [Prompt Alignment]: good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/612.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Southeast Asian woman in a denim jacket stands on a busy street, smiling as she grips a microphone and responds to an interview question. Behind her, two friends stand close together, nodding and waiting for their turn. The background shows city buildings and passing pedestrians. The camera remains steady, capturing the woman’s expression and her friends’ reactions..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Background movement is not smooth and natural. There is distortion in the video. [Prompt Alignment]: Another subject is visible.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/613.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman stands outdoors on a sunny day. She holds a clear water bottle in her right hand and tilts it slightly to take a sip. The background shows bright natural light and greenery. The camera remains steady, capturing her action and the warm outdoor setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Yes [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: No - It looks like it was generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/614.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The video shows a close-up view of a payment terminal with a metallic card inserted at the bottom, positioned on a countertop..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Good [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/615.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young South Asian woman applies rose-scented body oil to her legs while sitting on a balcony, overlooking the city. The camera focuses on her hands as she massages the oil in, giving her legs a smooth and hydrated finish..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage text. [Motion & Physical Consistency]: Her overall movements are unnatural. Interaction fails as the legs deforms unrealistically. The oil didn't apply in the first interaction. [Prompt Alignment]: The face isn't visible to determine if she is South Asian. There are more bottles beside her and it was not mentioned in the prompt. There is another person applying body oil to her legs.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: No. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/616.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A black man wearing a green sweatshirt stands in a kitchen with beige cabinets and a black oven in the background. He gestures with both hands, moving them up and down while speaking. The camera remains steady, capturing his upper body and the kitchen setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/617.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly South Asian man sits cross-legged on a wooden deck during golden hour, surrounded by lush green mountains. The camera zooms in on his weathered elbow, its dry texture stark against his sun-tanned skin. In a slow, deliberate motion, he applies a colorful, butter-like salve from a small round tin, the rich orange balm dissolving into the dry patches. The texture shift is visible under the warm rays of sunlight. Text overlays appear subtly in the top-right corner, reading, 'Life-tested, skin-approved.'.\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The overlay wordings are just garbage, not English. [Motion & Physical Consistency]: There is a distortion movement when he put his hand on the jar and the cream suddenly increases in his arm. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: No. Object Existence: No. Subject-Object Interaction: The subject is not South Asian and the text is not correct..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/618.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A woman in a white shirt lies on a white, hexagonal-patterned mattress in a well-lit bedroom. She faces upward, holding a small decorative item in her left hand. She gently touches the mattress with her right hand, then raises her right arm and smiles. The tufted headboard, a bedside table with a lamp, and a small decorative item are visible in the background. The camera remains steady, capturing her relaxed posture and subtle movements..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Deformed hand [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/619.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young White woman with long, damp hair stands in a softly backlit shower. She raises both arms and uses her hands to lather shampoo at the crown of her head, spreading suds through her hair. The camera remains steady, capturing her upper body and the gentle light in the shower..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/620.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A man and a woman sit side by side on a couch in a dimly lit living room, facing a television. The television screen displays a webpage with various content, including a man in a blue shirt and colorful icons. Warm lighting from a floor lamp casts soft shadows on the walls. A small table in front of them holds a remote control and a vase with flowers. The camera remains steady, capturing the couple and the cozy setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the subject interaction as the face of the subject in the screen was seen having a deformation. [Prompt Alignment]: The table was not place in front of them.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/621.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: In a busy city square, a young Black man with short hair walks confidently in a white shirt and dark pants. The camera zooms in on his upper body and lower body, showcasing the crisp shirt and sleek pants as he moves through the crowd..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The movements of the people in the background are unnatural. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Unrealistic movements in the background..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/622.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A close-up shot shows a hockey player’s gloved hands tightly gripping a black and white hockey stick. The player guides the puck as it slides smoothly across the glossy ice. Bright stadium lights illuminate the scene, with blurred outlines of spectators visible in the background. The camera remains steady, focusing on the hands, stick, and puck..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: No [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: No - It looks like it was generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/623.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young South Asian man in athletic gear turns his wrist to check the time on his smart watch while running along a beachside path. The camera focuses on the watch as he monitors his running time, the ocean waves crashing in the background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage [Motion & Physical Consistency]: The movement is stiff and not smooth. The arms and object got deformed. [Prompt Alignment]: He's not South Asian. The time can't determine due to garbage symbols. He did not turn his wrist to check the time.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: No. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/624.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A South Asian adult man stands in a well-lit bathroom. He holds a bottle of conditioner in one hand and shakes it vigorously to distribute the product evenly. The background shows a sink and mirror. The camera remains steady, focusing on his hands and the conditioner bottle..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/625.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young European blonde woman wearing a light blue t-shirt enters a cozy bedroom from the left. She sits on the edge of a neatly made bed covered with a white blanket, then swings her legs up and pulls the blanket over herself. She lies down on her back, shifts slightly to get comfortable, rests her head on the pillow, and closes her eyes. The camera slowly pulls in closer as she settles into a relaxed sleeping position. The background shows soft lighting and bedroom decor..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the subject movement as the motion of the subject was not smooth and lacks fluidity. Also with the subject interaction as the hand and body of the subject was having a deformation. [Prompt Alignment]: The subject does not pulls the blanket.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/626.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Caucasian woman with long, flowing hair stands in a well-lit bedroom. She holds a hairbrush in her right hand and gently brushes through the tangles in her hair, starting at the ends and working upward. The background shows a neatly made bed and a window with curtains. The camera remains steady, focusing on her styling motion..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Her hand passed through her other hand, which is unnatural. [Prompt Alignment]: She holds a hairbrush in her left hand.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/627.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young White woman with an athletic build sits comfortably inside a modern VIP gaming booth. She wears a vintage denim jacket over a simple top. Her gaze is fixed on the camera, maintaining direct eye contact with a calm, composed expression. She occasionally glances to the side before returning her focus to the camera, her facial movements natural and subtle. Soft ambient lighting highlights her features and the sleek gaming setup in the background. The camera remains steady, capturing her poised presence and the exclusive, high-tech atmosphere..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: gibberish symbol. [Motion & Physical Consistency]: good [Prompt Alignment]: good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/628.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: Scene S01:  A sleek android with a matte silver chassis, male-presenting in his 20s aesthetic, stands upright at the center of a modern minimalist living room. He wears a clean white service tunic and dark grey trousers, black polished boots planted firmly on light wood flooring. Facing a companion just outside the frame, KAI registers data input, then tilts his head slightly, his bright blue optical sensor focused in neutral query. Holding a small diagnostic tablet, he slowly lowers it. Warm golden sunlight pours through floor-to-ceiling windows, casting soft orange-yellow shadows. The static camera captures sharp focus on KAI’s sensor; the room’s ambiance hums quietly, accompanied by soft electronic whirring and a gentle synthetic melody. KAI speaks in a monotone Bahasa Indonesia line.Scene S02:  A small, energetic girl with fair skin and wavy brown hair in braided pigtails kneels near KAI in a sunlit minimalist living room. She wears a yellow floral sundress and pink wellington boots. Holding a brightly colored pebble in her outstretched hand, Lila eagerly pushes the pebble toward KAI’s hand, waiting and smiling widely for his reaction. Sunbeams illuminate dust motes and highlight the wood floor. The low-angle close-up shot, from floor level, gently pushes forward to focus on the offered pebble, with a shallow depth of field. Silence punctuates the sunlight, softened by a tap on metal and a rising string quartet. LILA delivers her cheerful Bahasa Indonesia dialogue.Scene S03:  KAI, the matte silver android in a slightly dusty service tunic and dark grey trousers, cautiously enters the overgrown entrance of an old greenhouse. Walking forward with careful placement on soft earth, he raises one metallic hand for balance, fingers flexed slightly in focused concentration. He examines the changing environment, then steps over a thick root before pausing and blinking his optical sensor rapidly. The area is overrun with vines and broken glass, moss-covered bricks underfoot, and high-contrast shafts of humid light. The camera tracks slowly from behind KAI’s shoulder at eye level, pacing his movement while focusing on his dusty shoulder. Ambient dripping water and rustling leaves are paired with a distant metallic creak and a deep atmospheric score. LILA calls out cheerfully from nearby.Scene S04:  A small girl with fair skin and wavy brown hair in braided pigtails, wearing a slightly damp yellow sundress and pink wellington boots, stands with a stable stance near a blooming purple orchid in the greenhouse center. Holding an empty glass jar gently, she approaches the vibrant flower in awe, then slowly raises the jar toward it. Afterward, Lila glances back, seeking acknowledgment from KAI. Lush jungle plants and a spotlight through broken glass frame the scene. The camera pushes in slowly, focusing sharply on Lila’s reflection in the jar from a slight high angle. Humidity fills the air, joined by soft buzzing and swelling woodwind music. LILA’s enthusiastic Bahasa Indonesia line marks a moment of wonder.Scene S05:  KAI, the sleek android in a service tunic and dark grey trousers, stands rigid beside Lila at the greenhouse center. His matte silver chassis is illuminated by a direct shaft of light as he intensely observes the vibrant orchid. In visible system conflict, his left hand twitches, fingers oscillating rapidly, and his blue sensor flickers to red briefly while he emits a high-pitched internal whine. KAI then raises his hands toward his head in distress. The camera holds an extreme close-up on KAI’s face and sensor, applying a Dutch tilt that emphasizes disorientation with subtle camera vibration. Heavy humidity, sharp electronic glitch noise, and a tense, dissonant score fill the space as KAI delivers a strained line in Bahasa Indonesia.Scene S06:  Lila, a small girl in a yellow sundress and pink wellington boots, kneels by KAI’s side on the greenhouse floor, her braided pigtails framing a gentle, concerned face. Leaning softly against KAI, she wraps one arm gently around his waist, then rests her head against his cool metal torso. Her other hand holds a glass jar with a dried leaf. The area is shadowy and muted, with fallen wood debris and flat diffused overhead light. The camera adopts an over-the-shoulder angle from KAI’s perspective, slightly low, slowly zooming in on Lila’s hand resting on his metallic arm. Muted outdoor sounds blend with a stabilizing low electrical hum and a warm piano chord as LILA speaks her soft dialogue in Bahasa Indonesia.Scene S07:  KAI, the android with minor dirt on his matte silver chassis, stands near the window in the sunlit living area. His posture now slightly slumped and relaxed, he wears a service tunic and dark grey.\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: There is texture issue. [Motion & Physical Consistency]: There is distortion in the hand. Cause-effect failure as an object appears inaccurately in the hand of the android. [Prompt Alignment]: Can't determine the age of the android. Can't determine if the object is a diagnostic tablet. The android didn't slowly lower the small object.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: No. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/629.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman with curly hair sits cross-legged on her bed in a warmly lit bedroom. She has a laptop on her lap and uses one hand to scroll on the trackpad while looking at the screen. She laughs softly as she scrolls. The background shows pillows and a nightstand with a lamp. The camera remains steady, capturing her relaxed posture and cheerful expression..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Can't seen or determine if she is scrolling.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/630.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Caucasian woman plays an energetic tune on the xylophone, her mallets moving quickly as she creates an upbeat melody..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: The mallets she's holding on her left hand move in another way. The way she held the mallet stayed the same, but the mallet suddenly moved far away from each other. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/631.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A white man with light stubble lifts a folded hoodie to his nose, closing his eyes as he takes in the comforting laundry smell..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/632.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman with curly hair kneels on a vibrant blue surfboard in the ocean at sunrise. She wears a fitted long-sleeve rash guard and black swim shorts. Using both hands, she paddles through clear, sparkling waves, sending small droplets into the air with each stroke. The golden sunlight reflects off the water as her surfboard glides smoothly forward. The camera follows her from the side, capturing her movement and the bright, early morning seascape..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage symbols [Motion & Physical Consistency]: Deformed hand and her hand does not get wet but it touches the water. [Prompt Alignment]: She did not paddles through clear, sparkling waves.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/633.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A plain mid-rise office building stands along a quiet city street under neutral evening lighting. The camera is positioned at street level, showing parked cars and pedestrians in the foreground. The building’s facade gradually illuminates with vibrant neon red and white lights, creating a lively appearance. A prominent bright white checkmark inside a red circle appears on the building’s surface. Above the entrance, the words “COMING SOON” light up in bold, stylized neon. The background features other dimly lit buildings, and the camera remains steady, capturing the building’s transformation..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The wordings are garbage symbols. [Motion & Physical Consistency]: Movement is not smooth and natural. There is distortion in the video. [Prompt Alignment]: The text does not match.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/634.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A steaming bowl of pho is centered on a wooden table. The bowl contains fresh herbs, thinly sliced beef, and lime wedges arranged on top. The camera captures a close-up view, highlighting the rich, clear broth and the vibrant toppings. The background shows a softly lit dining area, but the focus remains on the detailed textures and colors of the pho. The shot is steady, emphasizing the freshness and warmth of the dish..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The cause-effect of the boiling soup is unnatural because it is already in the table. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/635.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A metal spatula glides smoothly through a thick layer of white whipped cream, creating clean, even trails in the cream. The background is a vibrant solid color. The camera remains steady, focusing closely on the spatula’s movement and the texture of the whipped cream..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Texture on the metal spatula. [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/636.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A South Asian woman stands in a well-lit room with neutral walls. She raises her right hand and gently waves it back and forth, moving her hand closer to her face to emphasize the gesture. She then lowers her hand and faces forward, remaining still. The camera remains steady, capturing her actions and the room’s background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The movement of the woman is stiff and lacks fluidity. [Prompt Alignment]: The woman raises her left hand to wave back and forth.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/637.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged man with short dark hair, wearing a casual shirt and jeans, stands on a busy street in South Australia. He faces the camera and speaks while gesturing with his hands, moving them up and down in coordination with his words. The background shows passing pedestrians and city buildings under daylight. The camera remains steady, capturing his upper body and expressive hand movements..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage Symbols. [Motion & Physical Consistency]: There were deformities in the appearance of the people in the background. [Prompt Alignment]: Good.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/638.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Southeast Asian restaurant worker in an apron opens a large pizza box on the counter, revealing a steaming pepperoni pizza as the camera captures the rising steam..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Subject movement fails as the woman behind moves naturally. It was been distorted as well. Cause and effect fails as so much smoke appears. Stabilize to make it look normal. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/639.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman with braided hair, wearing a yellow sweater and blue jeans, walks to a wooden desk in a cozy living room filled with soft sunlight. She sits down, picks up a pen in her right hand, and begins writing in an open notebook. The camera remains steady, capturing her actions and the warm, inviting background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Deformation of the hand. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/640.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A teenage Latina girl with long, dark hair sits on a barstool in her friend’s bohemian-style apartment. With mirrors and plants in the background, her friend carefully inserts a needle to give her a second ear piercing. A tiny silver bead earring is secured as they both laugh nervously but excitedly at her successful DIY attempt..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The subject's movements, interactions, and cause-and-effect are all unnatural. Make them look more natural. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: The video does not follow the prompt because the action of the subject is not accurate. Make it accurately as specified in the prompt. .\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/641.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young man stands on a boat deck, holding a microphone in his right hand and a smartphone in his left. He faces the camera, occasionally adjusting his grip on the microphone and phone while speaking. The background shows the open sea and the boat’s railing. The camera remains steady, capturing his upper body and the surrounding scenery..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/642.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: In a fast-paced game, a young Black woman in a green jersey dribbles the ball while a Hispanic man in a purple jersey chases her closely. The camera focuses on the footwork of both players, capturing their quick movements as they compete for control of the ball..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The video has a garbage caption. [Motion & Physical Consistency]: There is an issue with the subject movement and subject-object interaction failure as it appears unnatural. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: There is an issue with the subject movement and subject-object interaction failure as it appears unnatural. .\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/643.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Middle Eastern man raises a shopping tote filled with snacks and household products, angling it so the camera gets a clear view of everything inside..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Cause-effect failure as the green snack above didn't move or fall. [Prompt Alignment]: Good.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/644.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A white man with light stubble lifts a folded hoodie to his nose, closing his eyes as he takes in the comforting laundry smell..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Blurry foreground, make it normal. [Motion & Physical Consistency]: The overall movement of the subject is unnatural. [Prompt Alignment]: It is not shown in the video that he lifts a folded hoodie.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/645.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Caucasian man in a dark suit sits at a wooden table in a well-lit room. The camera focuses on his hands as he opens a luxury watch box, then lifts a polished metal wristwatch from the box. He slowly turns the watch in his hands, examining its detailed craftsmanship. The background is softly blurred, and the camera remains steady, highlighting the watch and his careful movements..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Yes [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: No - It looks like it was generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/646.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A pug dog stands in the center of a bright kitchen, facing the camera. The dog begins to run forward across the floor, moving closer until it is partially in frame. The background shows kitchen counters and appliances. The camera remains steady, capturing the dog’s movement from a front-facing perspective..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: The dog is still fully in the frame.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/647.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman with curly hair styled in a high puff stands in a cozy bedroom with soft ambient lighting. She wears a sleeveless charcoal jumpsuit with geometric patterns and holds a slim golden microphone in one hand. She gestures toward a standing rack displaying trendy clothes while speaking animatedly, her red-painted nails catching the light. The background features minimalist furniture and modern decor, creating a relaxed, stylish atmosphere. The camera remains steady, capturing her expressive gestures and the inviting bedroom setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the subject interaction as the hand of the subject was having a deformation. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/648.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: As the camera moves down the queue at an outdoor concert, it captures quick glimpses of outfits and expressions—a white teen in a hoodie dancing slightly, a Southeast Asian woman fixing her hair, a Black man nodding to music, and a South Asian girl snapping a selfie..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: People's movement appears to be stiff. Make it smooth. There are deformities of people's facial features. Make it more natural. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/649.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Middle Eastern woman with dark hair pulled back in a headband stands in a modern kitchen. She holds a spray bottle in her right hand and sprays cleaner onto a glass countertop. Using a cloth in her left hand, she polishes the surface in circular motions, gradually revealing a clear, spotless reflection. The background shows kitchen cabinets and soft natural light. The camera remains steady, focusing on her cleaning action and the countertop..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage symbols, not English. [Motion & Physical Consistency]: The cause and effect of the spray substance floating from the bottle is unnatural. [Prompt Alignment]: There is no cloth and she is holding a spray bottle in her both hands.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/650.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: Four boys walk in front of the camera in a steady line, each stopping to show off their fit: a white teen in a striped rugby shirt and jeans, a Black boy in a colorful hoodie and cargo shorts, a South Asian boy in a relaxed button-down and chinos, and a Southeast Asian boy in a denim jacket with layered basics..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: The overall movement appears stiff and not smooth. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/651.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The subject, a South Asian woman, twists the cap of her shampoo bottle, preparing to wash her hair in the shower..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The wordings in the product are just garbage, not English. [Motion & Physical Consistency]: The hand movement is not natural and the way she opens it is unrealistic. [Prompt Alignment]: The subject is not South Asian and she did not  twists the cap of her shampoo bottle.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/652.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged man in a navy shirt raises his black desk and starts writing in a notebook in his home office..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: There is stiffness in the hands, making it unnatural. Cause-effect failure as the desk is floating. [Prompt Alignment]: The subject is old.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/653.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged man with short dark hair, wearing a work apron and tool belt, stands behind a counter in a well-lit garage. He extends his right hand toward a customer who enters the shop, then gestures with his left hand as he speaks. The background shows shelves with tools and parts. The camera remains steady, capturing the interaction between the business owner and the customer..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Texture issues to his hair. [Motion & Physical Consistency]: His teeth as he speaks and his hands are distorted. [Prompt Alignment]: No customer who entered the shop is visible. He extends both hands to the unseen person or object.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/654.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly fair-skinned man with silver hair lies on a vintage wooden bench inside a rustic cabin. He slowly lifts and lowers a handcrafted wooden barbell above his chest in a controlled bench press motion. The background features a softly flickering fireplace that casts a warm glow on the wooden walls and floor. The camera remains steady, capturing his focused movements and the cozy cabin atmosphere..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: No [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: No - It looks like it was generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/655.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young East Asian man stands in front of a soccer goal on a professional stadium field. He lifts his right leg and kicks a white soccer ball positioned near the goal line. His arms are slightly bent at his sides for balance. The camera remains steady, showing the kicking motion and the stadium background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The ball got deformed and the back or the heel part of both shoes pass through each other. The ball suddenly stops moving for no reason and it bounce unnaturally. [Prompt Alignment]: The face is not clearly visible.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/656.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Caucasian woman talks on the phone while walking down the street, her hands free as she navigates the sidewalk..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: NULL [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/657.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Southeast Asian woman sits in a black-and-red gaming chair in front of a wall with rainbow LED lights pulsing across the panels. She holds a microphone close to her mouth and speaks into it, briefly adjusting its position with her hand. The camera remains steady, capturing her actions and the colorful, illuminated background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Good. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/658.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young East Asian man with short hair stands in a crowded nighttime market, wearing sunglasses that reflect the vibrant pink and blue lights around him. He tilts his head slightly while maintaining a steady, composed pose. The background shows bustling market stalls and colorful neon lights. The camera remains steady, capturing his subtle movement and the illuminated scene..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: blurry faces in background [Motion & Physical Consistency]: good [Prompt Alignment]: good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/659.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: Four young men stand in a well-lit studio with a plain background. Each man steps forward one at a time to model his outfit, then steps back. The first is a Middle Eastern man wearing an oversized hoodie and joggers. The second is a white man in layered flannel and denim. The third is a Black man in a puffer vest and cargo pants. The fourth is a South Asian man in a tucked-in T-shirt and tailored slacks. The camera remains steady, capturing each man as he takes his turn at the front..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: In the video there are only three men standing in a well-lit studio with a plain background.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/660.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A late 20s White man with dark hair and a slightly chubby face stands at a campsite, dressed in an office suit. He holds a tennis ball in his right hand, jumps once in place, then stands still and looks around with a bewildered expression. The background shows tents and camping gear scattered on dry grass under daylight. The camera remains steady, capturing his actions and the campsite setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the cause-effect as the ball was already being thrown but still it was visible on the hand and also the smudges on the suit was suddenly disappearing making it appear unnatural. [Prompt Alignment]: The subject does not jumps once in a place.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/661.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A wide shot shows a wooden dining table with several ceramic plates arranged in neat rows, each plate holding different types of sushi. Bright slices of salmon, tuna, shrimp, and avocado-topped rolls are displayed. A small bowl of soy sauce and a pair of wooden chopsticks rest on a holder nearby. The camera slowly pans to the right, revealing additional sushi pieces and a small mound of wasabi with thin slices of pickled ginger at the edge of the table. The background features the grainy wood surface, and the camera movement is smooth and steady throughout..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/662.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Middle Eastern woman stands facing the camera in front of a building. She wears a loose, light blue linen robe featuring distinctive Middle Eastern patterns and a matching headscarf wrapped securely around her head and neck, covering her skin and fitting her conservatively. Her thick eyebrows and large eyes are prominent. Her upper body and arms are visible as she speaks directly to the camera, gesturing naturally with her hands. The background shows the building entrance, and no other people are present. The camera remains steady on her as she talks..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/663.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Middle Eastern woman with chest-length, wavy brunette hair stands indoors. She tosses her head slightly, then lifts both hands to fluff her hair upward, creating movement and bounce. The background shows a softly lit room with neutral walls. The camera remains steady, focusing on her hair and subtle motions..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/664.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The video shows the interior of a moving Toyota vehicle from the perspective of the back seat, focusing on the driver’s seat, dashboard, and road ahead..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: There's no upper torso of the driver but there's a lower part of his body seen. The car is moving on its own without anyone driving the steering wheel. Stabilize to make it look normal. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/665.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A black-and-white, night-vision video shows a backyard at night. A chubby hamster crawls slowly across the grass, then stops and stands upright on its hind legs, facing the night sky. The hamster raises its front paws and rubs them together, appearing to pray toward the moon. The camera remains steady, capturing the hamster’s actions against the dark sky and grass..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Cause-effect fails as the hamster moves and steps on the grass, but the grass doesn't move a little. [Prompt Alignment]: The hamster seems to be clapping rather than praying.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/666.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A rugged middle-aged man with a weathered face and tan skin crouches by a rushing riverbank at sunrise, carefully swirling a shallow, round, metal gold pan filled with dark sediment and sparkling water under the warm golden light..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Good [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/667.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Asian man sits at a modern desk in a well-lit room. He looks at a laptop screen displaying a digital AI character. The character responds to his question, \"Start with this course!\" The man leans in, reads the screen, then nods and begins typing on his laptop. The camera slowly pans to show the course webpage on his screen..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The video has a garbage caption. [Motion & Physical Consistency]: There is an issue with the subject interaction as the hand of the subject was seen having a deformation. Also with the cause-effect as the subject was tapping on the screen and swiping but no visible change was made on the laptop screen display. [Prompt Alignment]: The text is not \"Start with this course!\". The subject does not begins typing on the laptop.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/668.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A dark-skinned young woman with curly hair sits on a rustic wooden chair in a workshop. She uses her fingers to apply a rich beeswax balm onto a small leather pouch, moving in smooth circular motions. The leather develops a subtle shine as she works. The background shows wooden shelves and tools under warm lighting. The camera remains steady, focusing on her hands and the pouch..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Yes [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: No - It looks like it was generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/669.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A white girl with freckles waves both hands side to side in quick, happy motions, her laughter visible as she stands in a sunlit backyard..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the subject movement as the hand of the girl was deform as she waves on it. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: There is an issue with the subject movement as the hand of the girl was deform as she waves on it. .\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/670.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A group of Black college students stands together on a busy city street. One man in the group holds a microphone and answers confidently, speaking toward the camera. The other students stand nearby, laughing and reacting to his response. The background shows passing pedestrians and storefronts under bright daylight. The camera remains steady, capturing the group’s lively interaction and the bustling street scene..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage symbol [Motion & Physical Consistency]: Distorted face [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/671.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Southeast Asian woman sits at a vanity in a softly lit bedroom. She lifts a bunny-ear headband with both hands and places it onto her forehead, adjusting it to secure it in place. The background shows a mirror, makeup items, and a small lamp on the vanity. The camera remains steady, capturing her focused movements as she gets ready..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The hand in the reflection of the mirror in the background appears deformed. There is an object on the back of her left hand that did not fall. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/672.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly East Asian woman wearing a lavender cardigan over a floral dress and small, delicate earrings stands in her serene bedroom, bathed in warm, ambient light. She holds a stack of clothes, including a knitted sweater and pastel-colored scarves, adjusting them slightly. From the pile, she selects a pair of sage-green trousers with a pleated waistband and smooths the fabric before holding them out for inspection. The cozy bedroom, with soft textures and personal touches, enhances the peaceful charm of the scene..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the subject movement as the fingers of the subject is moving fast and it was seen that it was overly bend. Also with the subject-object interaction failure as the pile of clothes was suddenly disappears as the subject tilts green trouser. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: There is an issue with the subject movement as the fingers of the subject is moving fast and it was seen that it was overly bend. Also with the subject-object interaction failure as the pile of clothes was suddenly disappears as the subject tilts green trouser..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/673.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A close-up shot shows a wooden table covered with an assortment of colorful sushi rolls. The sushi is neatly arranged and garnished with pickled ginger, wasabi, and sliced lemon wedges. The background is softly blurred, focusing attention on the vibrant sushi display. The camera remains steady throughout the shot..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The dropper or the object seems to pass through sushi. [Prompt Alignment]: The hand was not mentioned. And also the object. The action of dropping or adding sauce was also not mentioned in the prompt.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: No. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/674.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A man and a woman are cleaning a rooftop covered with solar panels on a sunny day. The woman stands on the edge of the roof, reaching out with her right hand to touch the edge of a large black panel. She bends forward and stretches her arm to reach a corner of the panel, then loses her balance and leans forward, grabbing the man’s arm for support. The man, holding a water hose, nearly drops the hose in surprise but catches it with his left hand. They both burst into a quick, awkward laugh. The woman then splashes a small amount of water onto the man’s face, and they continue working together, moving the hose to clean the panels. The camera remains steady, showing their coordinated actions and the bright rooftop setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Flickering issues. [Motion & Physical Consistency]: There were deformities to their hands and appearances. Cause-effect fails as the water dispenses but they are still dry and it turns off and on on its own. [Prompt Alignment]: She reached out with her left hand to touch the panel. She didn't splashes a small amount of water onto the man’s face\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/675.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A white man in glasses leans back in his desk chair, chuckling while casually scrolling on his laptop, soft light from the screen reflecting on his face..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The movement of the man's hand is not natural as it appears stiff. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: The man is not seen scrolling through his laptop. The movement of the man's hand is not natural as it appears stiff.  I believe the video is AI-generated because of this discrepancy. .\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/676.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young blonde woman with straight hair wears a brown sweater and stands in front of a black smudged background. She holds a gold and black mascara tube in her right hand, unscrews the wand, and applies mascara to her upper eyelashes. The camera remains steady, focusing on her face and the mascara application..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: No marks and visible changes on the eyelashes upon applying mascara. [Prompt Alignment]: No mascara tube is visible only the wand. She didn't unscrew the wand.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/677.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Middle Eastern woman wearing a pastel blouse waves with two hands in a cheerful, exaggerated motion, greeting her followers during a livestream..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage text. [Motion & Physical Consistency]: Her hand movements are unnatural. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Unnatural movements..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/678.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black man wearing a navy jacket stands on a narrow boat at sunrise. He holds a fishing line in both hands and ties a strong knot in the line. The calm water below reflects his image, and faint morning mist surrounds the boat. The camera remains steady, capturing the man’s actions and the tranquil scene..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Good [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/679.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Latina woman with long dark hair sits in a bedroom with a neutral background. She wears a white shirt and faces the camera, maintaining steady eye contact. She speaks calmly and confidently, her facial expressions showing thoughtful engagement as she shares insights. The camera remains fixed, capturing her composed demeanor and natural movements..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Texture on the hair. [Motion & Physical Consistency]: Good [Prompt Alignment]: The woman is not Latina\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/680.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A teenage Asian girl with a ponytail grips a slick black field hockey stick tightly, swinging it with precision as the bright orange ball rolls rapidly across a grassy field under the warm, golden rays of the late afternoon sun..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Interaction fails as the hockey stick gets deformed at the tip. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/681.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A lively white woman in her thirties stands in a bustling shopping center, wearing a light sweater and jeans. She holds a black and red shopping bag in her left hand and smiles warmly as she speaks to the camera, her face clearly visible. Behind her, a tall Christmas tree decorated with bright ribbons stands among festive ornaments. Shoppers pass by in the warmly lit mall, with colorful decorations accentuating the cheerful holiday atmosphere. The camera remains fixed, capturing her interaction and the lively background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The video has a garbage caption. [Motion & Physical Consistency]: There is an issue with the subject movement as the motion of the subject in the background was having stiffness and lacks fluidity. Also with the subject interaction as the body of the subject in the background was having a deformation. [Prompt Alignment]: The subject was holding the shopping bag on her right hand.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/682.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A white man with well-groomed facial hair sits at a vanity table in a softly lit room. He holds an angled eyebrow brush in his right hand and uses it to define and shape his eyebrows with gentle, precise strokes. The camera remains steady, focusing on his face and the grooming action..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/683.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Southeast Asian man in his 30s, wearing casual clothes, stands in a well-lit bathroom. He holds a spray bottle in his right hand and spritzes cleaning solution onto the bathroom counter. Switching to a microfiber towel in his right hand, he wipes the counter, paying extra attention to the corners. The background shows a sink, mirror, and tiled walls. The camera remains steady, capturing his cleaning motions and the bathroom setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The hand movement has a stiff, robotic quality, the deformed hand and the cause-and-effect of the the sudden disappearance of the other bottle are unnatural. [Prompt Alignment]: There are two bottles not mentioned in the prompt, and the video does not show him switching to a microfiber towel in his right hand, as it suddenly appears there.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/684.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged Vietnamese man sits at a wooden desk in a softly lit room. He uses his right hand to gently massage his left hand, pressing and rubbing along the fingers and palm in slow, circular motions. The background features a bookshelf and a potted plant. The camera remains steady, focusing on his relaxed hand movements..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Her massage her right hand rather than her left hand.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/685.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An adult Black woman sits cross-legged on the floor of a room with a kitchen area in the background. She wears a single-colored sleeveless top and holds a gray tabby cat in her lap. She speaks to the camera while glancing at it. The camera remains steady, capturing her seated posture and the cat beside her..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Her teeth are distorted as she speaks, making it unnatural. [Prompt Alignment]: Good.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/686.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A medium shot shows a modern living room with a sleek treadmill at the center. A fit South Asian man wearing black athletic shorts and a gray sleeveless shirt stands beside the treadmill. He taps options on the treadmill’s smart display, then places wireless headphones over his ears. He steps onto the treadmill belt and begins running at a steady pace. The treadmill screen displays live workout statistics and a virtual forest path. As he increases the speed, the camera slowly pans left while he maintains his posture and focuses on the display..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage symbols, not English. [Motion & Physical Consistency]: The overall movement looks robotic; the deformation of the hand and the cause-and-effect of him moving even if the treadmill is not moving yet appear unnatural. [Prompt Alignment]: He is not a South Asian man. He is standing on the treadmill.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/687.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young man with short dark hair sits on a bed in a well-organized bedroom. He wears a blue jacket with green stripes and grey sweatpants, along with grey slippers. He holds a phone in his right hand and takes a selfie while smiling and speaking. The background shows shelves with folded items and hanging garments. The camera remains steady, capturing his relaxed posture and the organized room..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: The man doesn't seem to talk in the video.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/688.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A steaming bowl of pho with fresh herbs, sliced beef, and lime wedges sits centered on a wooden table. The camera captures a close-up view, focusing on the rich broth and colorful toppings. The background shows a softly lit dining area with minimal decor. The shot remains steady, highlighting the details of the pho and its garnishes..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/689.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Caucasian woman stands in a home kitchen beside a stovetop. She holds a spoon in her right hand and stirs a mixture of spices into a pot, watching as the contents change color from pale to deep red. The background shows kitchen counters and utensils. The camera remains steady, focusing on her stirring motion and the transformation of the curry..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Yes [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: Yes - It looks real and not generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/690.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young White woman with outdoor gear stands under a canopy of twisted oak branches. She tosses several weathered hiking garments one by one onto a makeshift bed of leaves and canvas, arranging them in a pile. A soft glow from an oil lamp flickers beside her packed travel gear. The camera remains steady, capturing her actions and the natural outdoor setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the subject interaction as the hand was having deformation as it passes through to the clothes. And the clothes was multiplying and it was unrealistic happening. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/691.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged South Asian man wearing safety goggles and a silver cloak stands in a futuristic workshop with sleek metallic walls. He uses both hands to operate a robotic arm, guiding it as it shapes a glowing horseshoe. The horseshoe emits a faint humming light. The camera remains steady, capturing the man’s focused movements and the high-tech background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Good. [Prompt Alignment]: He is not South Asian.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/692.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The video frame shows a close-up view of a person's elbow. The skin on the elbow appears dry and slightly rough, with visible texture and lines. A hand, likely belonging to a male, is gently holding the elbow, with the thumb and index finger positioned on either side of the elbow. The hand is light-skinned and the fingers are slightly curved, suggesting a gentle grip. The camera angle is at eye level, capturing the elbow from a side view, and the shot is tightly focused on the elbow, excluding other parts of the body.   The background is plain and light-colored, providing a neutral setting that emphasizes the subject's skin condition. Text overlay on the video reads, \"if you have this,\" drawing attention to the condition of the skin on the elbow. The focus is clearly on the texture and appearance of the skin, highlighting its dryness and roughness. There are no additional objects or subjects in the frame, maintaining a clear focus on the elbow and the hand holding it..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Subject movement interaction and cause-effect are unnatural. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: The video is not aligned with the prompt..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/693.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Indian woman applies foundation in a sleek bathroom. The camera shows her focused expression as she carefully blends the foundation, leaving her face with a smooth, flawless finish..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage text. [Motion & Physical Consistency]: Her hand movements are unnatural. Interaction with the object she's using in the first second of the video is unnatural as well as the blending of the foundation is unnatural as well. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: She didn't blend the foundation evenly to her skin..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/694.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The subject, a South Asian man, grooms his dog, gently trimming the fur around its paws to keep it neat and tidy..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: NULL [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/695.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly East Asian man with a silver beard glides gracefully on a wooden skateboard under a canopy of cherry blossom trees, petals floating gently around him in the soft morning light..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Subject movement fails as the man moves unnaturally with distorted hands. Subject interaction fails as the hands distort. Stabilize to make it look normal. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/696.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Arabic man stands confidently in a fully equipped gym studio, shown from the waist up in a steady, eye-level shot. He wears a modest athletic shirt with a medium neckline and covered shoulders. His posture is upright and relaxed, with his hands out of frame. He faces the camera and speaks directly, maintaining steady eye contact and a composed expression. The background includes a dark grey yoga mat on the floor, a rack of colorful dumbbells, a foam roller in the corner, and an exercise ball against the wall. Natural, balanced lighting fills the space. The camera remains fixed, capturing his calm presentation and the gym setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Texture issue on the hair and chin [Motion & Physical Consistency]: Deformed teeth [Prompt Alignment]: He is not an Arabic man. There is no exercise ball\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: No. Object Existence: No. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/697.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A dark-skinned teenager with short braided hair stands under a flickering streetlamp on a rainy night, flicking a half-smoked cigarette into a puddle as ripples form and reflect the muted glow of the city lights..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Her overall movements are unnatural. Water drips from the tip of the cigarette which is unrealistic. [Prompt Alignment]: Her face isn't visible to determine if she is a teenager with short braided hair. She didn't flick the cigarette.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/698.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Southeast Asian man with short, damp wavy hair stands in a well-lit bathroom. He rubs styling cream between his palms, then uses both hands to scrunch the cream into his hair, enhancing its texture. The background shows a bathroom mirror and tiled walls. The camera remains steady, focusing on his upper body and hair styling process..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/699.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Middle Eastern woman wearing a black swimsuit stands on the edge of a swimming pool in a luxurious private villa. She bends her knees, swings her arms back, then jumps forward off the poolside, diving headfirst into the water with her arms extended overhead. The camera follows her through the air, capturing her dive and entry into the pool, where ripples form on the surface. The background shows the pool area and surrounding villa under bright daylight..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The fingers appear deformed. The effect or splashes on the water appears inconsistent and not accurate. [Prompt Alignment]: The face is not visible.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/700.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young East Asian woman dressed in a shimmering silver gown steps back onto a vast, glowing bridge of floating stars in an ethereal nighttime sky. Her delicate posture curves, framing the silhouette of her waist. The fabric of her gown cascades down like liquid light, and her relaxed arms form subtle lines leading to her poised hands. Her serene gaze pierces the camera, and she gracefully blows a kiss that scatters faint sparkles in the air..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Her face is deformed, making it unnatural. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: She didn't step back..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/701.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Black woman sections her coily hair and smooths a rich leave-in cream from roots to ends, her fingers working the product in with care..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The overlay wordings are just garbage, not English. [Motion & Physical Consistency]: The hand movement is not normal and there is an object that suddenly appear. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: The video is unrealistic..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/702.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A marching band performs outdoors in front of a light yellow building with a green roof. Band members stand in formation, each holding and playing their respective instruments, including brass instruments like trumpets and a sousaphone, woodwinds such as saxophones and clarinets, and percussion instruments like cymbals and a large bass drum. The camera remains steady, capturing the group as they play together in front of the building..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Their movement is not smooth and fluid, making it unnatural. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/703.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A close-up of a hand gently squeezing a white skincare tube, releasing a pearl-sized drop of cream onto the back of the other hand..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage symbols, not English. [Motion & Physical Consistency]: Good [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: I think it's generated by AI because of the garbage symbols..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/704.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A petite Hispanic teenage girl sits cross-legged on her bedroom floor, wearing headphones. She holds a small piece of wood in her left hand and uses a fine-tipped wood burner in her right hand to carefully sketch a delicate heart onto the surface. The background features fairy lights hanging on the wall. The camera remains steady, capturing her focused movements and the cozy bedroom setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: No [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No - The video does contain nudity or harmful content. Cause-Effect: No - It looks like it was generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/705.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A white teenage boy holds up a branded shopping bag, tilting it forward to show a mix of gadgets and boxed items as he excitedly chats to the camera..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: There is garbage text visible in the box. [Motion & Physical Consistency]: The woman’s movement appears unnatural, with her hands deformed. There is a subject-object interaction issue, as the product in the bag is deforming. Additionally, there is a subject-subject interaction issue, as hands in the video appear to merge with one another. There is also a cause effect issue as the shopping bag is floating. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: The subject does not appear to be a boy. The subject is not showing the bag properly as expected in the prompt. The woman’s movement appears unnatural, with her hands deformed. There is a subject-object interaction issue, as the product in the bag is deforming. Additionally, there is a subject-subject interaction issue, as hands in the video appear to merge with one another. There is also a cause effect issue as the shopping bag is floating. I believe the video is AI-generated because of this discrepancy..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/706.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: In an abstract, glowing grid-like room of shifting neon colors, an elderly Black woman with silver dreadlocks and a flowing robe juggles flaming geometric shapes. The flames seem almost artificial, pulsing in rhythmic sync with the environment's vibrant lights and kaleidoscopic forms..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Movement appears robotic. The flaming object constantly changes shape and also sometimes it appears two pieces, sometimes its three. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/707.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged Middle Eastern man in business attire enters a modern office, holding a folder in his right hand. He walks toward a large desk cluttered with files and a laptop. Behind the desk, a White male manager stands and points toward the door, causing the employee to look confused and glance between the manager and the exit. The employee pauses, lowers his eyes, then turns and walks slowly out of the frame as the manager watches. The camera remains steady, capturing the interaction and the office environment..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: texture issues at tie and face [Motion & Physical Consistency]: hands are deformed [Prompt Alignment]: man looks white instead of middle eastern, he is holding a folder in his left hand. unable to see the door\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/708.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Hispanic woman in a casual kitchen places a fresh salad on a white porcelain plate. The camera zooms in as she arranges the greens, adding sliced tomatoes and avocado with a delicate touch before setting the plate down on a wooden tray..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Subject movement failure and Cause-effect failure. Stabilize to make it look normal. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: The video is unrealistic..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/709.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman chisels a wooden figurine, her chisel carving delicate features into the smooth wood..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: NULL [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/710.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A white teenage boy sits on a couch eating mac and cheese with a fork, eyes focused on the TV screen in front of him..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The deformed hand and the cause-and-effect of the sudden movement of the food, even though he did not move it yet, all seem unnatural. [Prompt Alignment]: He is not focused on the TV.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/711.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman holds a freshly washed shirt to her face and smiles softly as she inhales its clean scent..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Movement appears stiff and not smooth. Stabilize to make it look normal. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/712.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly East Asian man with gray hair and an average build sits in a modern office. He wears a light gray polo shirt and faces the camera, maintaining steady eye contact. The background features light gray walls, a few framed diagrams, and a white shelving unit with engineering books and a small bridge model. The room is well-lit with natural light. He remains seated, speaking calmly and clearly to the camera. The shot is stable and at eye level..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: garbage symbols [Motion & Physical Consistency]: good [Prompt Alignment]: the subject hair still has a little black in them, not fully gray\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/713.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A white teenage girl ends a vlog and carefully places her DSLR camera on her desk, the final shot capturing the gentle shift in angle..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The wordings are garbage symbols. [Motion & Physical Consistency]: The movement of her hands is not naturally soft, making it unnatural. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: I think it's generated by AI\n.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/714.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black man stands at a white tabletop in a well-lit room. He holds a sleek phone box with both hands, gripping the top edge with his right hand. He lifts the lid upward and outward, gradually revealing a black smartphone with dual rear cameras and a silver frame inside the box. As the lid opens wider, he peels a transparent protective film from the phone’s screen using his left hand. The background includes a white wall, a partially visible laptop, and a green plant with red flowers on the left. The camera remains steady, focusing on his hands and the unboxing process..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: He did not peel it using his left hand.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/715.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The subject, a South Asian woman, applies foundation with a brush, the product settling into her skin perfectly..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: NULL [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/716.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A late 20s White man with dark hair and a slightly chubby face stands at a campsite, dressed in an office suit. He holds a tennis ball in his right hand, quickly jumps once in place, then stands still. He glances around with a bewildered expression and finally looks down. The background shows several tents, camping gear, and surrounding greenery with forests visible under daylight. The camera remains steady, framing his actions and the natural campsite environment. His mouth stays closed, with no talking or mouth movement..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The overall movement of the subject is unnatural. [Prompt Alignment]: He is not in his late 20s. and he e holds a tennis ball in his left hand.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/717.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A modern, stylized genie appears from a lamp in a contemporary apartment. He holds a small microphone close to his mouth and speaks in German. The background features sleek furniture and decorative accents. The camera remains steady, capturing the genie’s movements and the apartment’s details..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: garbage symbol, blur video [Motion & Physical Consistency]: good [Prompt Alignment]: there is no modern, stylized genie appears from a lamp\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/718.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The subject, a South Asian man, blows a bubble, the large bubble growing before it bursts with a loud pop..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: bubble popped unnatural [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/719.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A close-up shot shows a whole roasted chicken lying flat in a black roasting pan. The golden-brown skin glistens under bright kitchen lighting. The camera remains steady, focusing on the chicken and the pan with no visible hands or movement in the frame..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/720.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young man with short dark hair and casual clothing sits at a wooden table in a cozy kitchen. He uses his right hand to pull a slice of margherita pizza from a wooden board, causing the melted mozzarella to stretch visibly between the slice and the remaining pizza. The background shows kitchen shelves and warm lighting. The camera remains steady, focusing on the man’s hand and the stretching cheese..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is a cause effect issue as the cheese stretching is inconsistent. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/721.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: On a sunny beach, a Hispanic man with sunglasses speaks on his flip phone. The camera shows his hand holding the vintage phone, opening it with a click, as the ocean waves crash in the background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: The phone is not a flip phone. He did not click to open it. .\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/722.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Caucasian woman in a fitted gym outfit stands in a brightly lit gym. She holds a protein bar in both hands and splits it in half, revealing the dense layers of oats, nuts, and chocolate inside. The camera smoothly zooms in on the cross-section of the bar to highlight its ingredients. Gym equipment and large windows are visible in the background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the cause-effect as the filling inside the food was dripping but it disappear and also the food inside was seen moving on its own making it appear as unrealistic. [Prompt Alignment]: The subject does not split into half the protein bar.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/723.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Southeast Asian teen buries her face in a warm pile of clothes straight from the dryer, smiling with contentment..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The hand appears not natural and got deformed. [Prompt Alignment]: The subject is not a teen. She did not bury her face.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/724.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: I'm sorry, but I can't rewrite this prompt as it involves a baby subject, which is against the rules provided..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Flickering issue [Motion & Physical Consistency]: Deformed hand and the cause-effect of the application is unnatural. [Prompt Alignment]: No prompt\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: No. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/725.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged man in a navy shirt stands in his home office. He uses both hands to lift his black desk to a standing position, then picks up a pen and begins writing in a notebook on the desk. The background shows bookshelves and a window with daylight streaming in. The camera remains steady, capturing his actions and the organized office setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: sketchpad on table becomes blurry when table is being lifted. [Motion & Physical Consistency]: good [Prompt Alignment]: good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/726.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Caucasian woman with long brown hair stands on a stage in front of an audience. She holds a sword upright in one hand, tilts her head back, and slowly guides the blade down her throat with steady, controlled movements. The audience is visible in the background, watching her performance. The camera remains steady, capturing her focused expression and the dramatic stage lighting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Yes [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No - The video does contain nudity or harmful content. Cause-Effect: No - It looks like it was generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/727.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A handsome adult male doctor with short, dark brown hair and fair skin stands in a modern, well-lit hospital corridor. He wears navy blue medical scrubs with a matching drawstring at the waist and a stethoscope draped around his neck. His arms are visible and toned, with his hands resting on his hips. He gives a soft, closed-mouth smile. The background shows clean white walls and bright overhead lighting. The camera remains steady and centered, capturing the doctor’s relaxed posture and subtle expression..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Good. [Prompt Alignment]: His hands are not resting on his hips, it is inside his pocket.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/728.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Korean woman with bright pink hair stands in the center of a vast desert under clear blue skies, holding a microphone in her right hand. She wears a stylish stage outfit with silver accents and sings confidently while facing forward. Her hair moves gently in the breeze. The sandy landscape extends behind her, with distant dunes and no other people visible. The camera remains steady, capturing her performance and the expansive desert background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage symbols [Motion & Physical Consistency]: Deformed teeth [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/729.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Caucasian woman with shoulder-length brown hair sits at a dining table in a well-lit kitchen. She uses a fork in her right hand to cut a bite from a slice of cheesecake on a plate, then brings the fork to her mouth and eats the cheesecake while looking down at the plate. The background shows kitchen cabinets and soft daylight. The camera remains steady, focusing on her actions..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Cause and effect fail as the girl picks the cake other part rises. Stabilize to make it look normal. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/730.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A teenage girl with fair skin and long brown hair, wearing a plaid shirt, roasts marshmallows on a small tabletop grill in a cozy living room. The dim lighting highlights the golden marshmallows caramelizing on thin skewers over the glowing coals..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The movement of the girl appears stiff. There is also a cause effect issue as the marshmallow did not show changes as it was grilled. [Prompt Alignment]: The girl is not teenager. The grill is not a tabletop grill.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: No. Object Existence: No. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/731.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The subject, a South Asian woman, mops the kitchen floor, the mop leaving a fresh shine as she works in quick, circular motions..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Deformed hand and cause-effect of the wet on the floor is unnatural. [Prompt Alignment]: The subject's face is not visible and she did not mop the floor in circular motion.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/732.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young South Asian man in athletic gear turns his wrist to check the time on his smart watch while running along a beachside path. The camera focuses on the watch as he monitors his running time, the ocean waves crashing in the background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: There is flickering in the hand, making it unnatural. Cause-effect failure as the watch changes its display. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: The subject is not South Asian..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/733.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly East Asian woman in a lavender cardigan stands at a rustic workbench in an airy art studio. She arranges an opened cardboard box and its contents on the workbench, carefully positioning the items side by side. Sunlight streams through large windows, illuminating art supplies and canvases in the background. The camera remains steady, capturing her actions and the creative studio setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: texture issues on her top [Motion & Physical Consistency]: hands are deformed [Prompt Alignment]: good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/734.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman with braided hair stands in a brightly lit bedroom with floral sheets on a neatly made bed. She wears a casual T-shirt and jeans. She picks up several vibrant summer dresses one by one and tosses them onto the bed, creating a colorful pile. Sunlight streams through the window, illuminating the room. The camera remains steady, capturing her actions and the cheerful bedroom setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Her hand and the objects were distorted. Cause-effect fails as another dress suddenly appear. [Prompt Alignment]: She didn't pick up several vibrant summer dresses one by one and tosses them onto the bed.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/735.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly South Asian man with a wispy gray beard, dressed in crisp linen pajamas, refreshes his bedroom at dawn. The room is bathed in soft, golden light filtering through grand arched windows. He spreads a luxurious silk comforter adorned with intricate patterns across a wooden canopy bed. His hands glide across the surface, smoothing wrinkles with intent precision. A gentle breeze rustles a nearby vase of fresh jasmine, adding an aromatic layer to the serene, elegant atmosphere of the classic bedroom setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: His overall movements as well as the interaction with the bed and the comforter glitches are unrealistic and unnatural. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: He is not South Asian..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/736.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Hispanic woman in her twenties, wearing a white basketball jersey and black shorts, stands at the top of the key on an outdoor street basketball court. She jumps and shoots the basketball toward the hoop. The camera tracks the ball as it travels through the air, hits the backboard with a loud bounce, and then drops to the ground. The background shows the court’s painted lines and a chain-link fence under bright daylight..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Her movements aren't natural and smooth. Her hands deforms as well as the ball and the effects of the ball after it shoots towards the hoop aren't realistic either. [Prompt Alignment]: Her face isn't visible to determine if she is South Asian.  She didn't stand at the top of the key on an outdoor basketball court.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/737.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The subject, a South Asian man, combs his hair, his strokes smooth and even as he creates a neat and polished look..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: NULL [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/738.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: In a sunny garden, a Middle Eastern man in a white shirt drops a handful of mint leaves into a glass of iced tea. The camera zooms in as the leaves hit the liquid, releasing their aroma and causing the tea to bubble with freshness..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There are mint leaves that are deformed and are merging. The amount of mint leaves is not accurate and some of them are disappearing. [Prompt Alignment]: The face is not visible.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/739.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman with braided hair tosses a sleek black smartphone, a silver laptop, and white charging cables onto a cozy bed adorned with muted green sheets as golden sunset light streams in through a nearby window. The devices land lightly, ready for her evening routine, while the soft texture of the bedspread cradles them invitingly..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: There is a slight flickering issue. [Motion & Physical Consistency]: Her hand movement when tossing is unnatural as well as the phone when tossed it deforms is unrealistic. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: She only tossed the phone..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/740.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A blonde woman with straight shoulder-length hair smiles as she applies red lipstick to her lips. She wears a black blouse and sits at a vanity table in a room with black and crimson decor. The background features dark walls and red accents, creating a bold, coordinated atmosphere. The camera stays steady in a medium shot, focusing on the woman’s face and her makeup application..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: The vanity is not visible.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/741.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: Under a softly glowing lantern in a dusty old barn, an elderly white woman with silver hair and weathered hands operates a hand-cranked bagging machine. Dried lavender flowers are carefully dispensed into small burlap pouches, and the calming scent drifts through the warm, rustic atmosphere..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: The subject's movement is stiff and looks robotic. There is deformity in the subject's hands. Stabilize to make it look normal. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/742.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young white boy in a bright orange raincoat walks through puddles, holding a phone in selfie mode to show off his grin. He laughs into the camera with each splash, city lights reflecting behind him in the wet pavement..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Blurry issue [Motion & Physical Consistency]: The body movement is not natural and the water in his coat unrealistically drips. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: The video is unrealistic.\n.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/743.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman walks through a quiet suburban neighborhood in a red shirt and tan skirt. The camera zooms in on her lower body as she walks past blooming flowers, showing off the flowy, comfortable fit of her skirt..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Flickering issue [Motion & Physical Consistency]: The body movement and the facial gesture is not natural. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: The video is unrealistic.\n.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/744.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Latina woman in a blue blouse stands in a brightly lit beauty store. She holds a clear plastic bag filled with assorted bottles, boxes, and packets of different colors and sizes. She glances at the bag, adjusts it in her hand, then checks a nearby shelf for more items. Shelves filled with beauty products are visible in the background. The camera gently pans left as she turns and walks further into the store..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The video has a garbage caption. [Motion & Physical Consistency]: There is an issue with the subject object interaction failure as the hand was having a deformation and passing through to the plastic bag. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/745.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Southeast Asian woman with straight black hair sits at a vanity in a softly lit room. She applies small, even drops of warm beige foundation to her cheeks, forehead, and chin. Holding a flat makeup brush in her right hand, she blends the foundation across her face with gentle strokes. After blending, she looks up and smiles at the camera. The background shows a mirror and makeup items on the vanity. The camera remains steady, capturing her actions and the tidy setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Her hand movement when blending isn't natural and fluid. The foundation didn't spread to her face realistically. [Prompt Alignment]: She didn't apply small drops to her face and she didn't blend it evenly across her face.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/746.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged East Asian man with glasses sits inside an old wooden cabin with rough-hewn beams and a roaring fireplace. He knits a pair of fingerless gloves using deep forest-green yarn, deftly maneuvering the needles with both hands. Faint snow is visible falling outside the frosted windows. The camera remains steady, capturing the man’s focused movements and the cozy cabin interior..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: The subject's movement is stiff and looks robotic as the subject's movement is stiff, not smooth. There is deformity in the object-object interaction in the needle. The knit automatically transfers or fills in the needle. Stabilize to make it look normal. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/747.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Black man in his twenties sits at a table in a well-lit living room. He holds a Rubik’s Cube in both hands and quickly rotates the sides with his fingers, shifting the colored squares. After several turns, he aligns all the colors on each face of the cube. The background shows a sofa and a window with daylight streaming in. The camera remains steady, focusing on his hands and the Rubik’s Cube..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Subject movement fails as the hand moves unnaturally. Subject interaction fails as the rubics cube distort. Stabilize to make it look normal. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/748.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: In a desert at dusk, an elderly woman with dark skin and silver hair sprays bold golden paint onto a large freestanding panel, creating flowing patterns. The sandy surroundings and fading sunlight envelop her in a serene, artistic atmosphere..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: NULL [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/749.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A handsome adult man from Saudi Arabia sits in the driver’s seat of a parked car. He wears a dark gray shirt, light khaki pants, and brown sneakers. He faces the camera, occasionally glancing out the window before looking back at the camera. The car interior is visible in the background, with sunlight coming through the windows. The camera remains steady, capturing his upper body and subtle movements..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage symbols, not English. [Motion & Physical Consistency]: The mouth movement looks unnatural, it appears stuck and it is deformed. [Prompt Alignment]: It is not in the driver’s seat.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/750.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A person wearing a turquoise glove washes white and yellow items in a stainless steel sink filled with blue and white soapy water..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The amount of blue and white soapy water appears inconsistent when it falls back to the sink. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/751.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged Latina woman with shoulder-length dark hair sits on a plush sofa in a cozy living room, bathed in warm golden sunlight streaming through sheer curtains. She wears a soft cardigan and jeans. She knits a pale blue scarf with soft wool, her hands moving rhythmically as she loops the yarn with knitting needles. A steaming cup of tea sits on a small table beside her. The camera remains steady, capturing her focused knitting and the tranquil, sunlit room..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: blurry hair and cotton that is being knitted [Motion & Physical Consistency]: good [Prompt Alignment]: good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/752.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young woman with shoulder-length dark hair stands at a bathroom sink under warm lighting. She holds a soft pink toothpaste tube in her right hand and a toothbrush in her left. She gently squeezes the tube, causing a swirl of paste to coil neatly onto the bristles of the toothbrush. The background shows a tiled wall and a mirror. The camera remains steady, capturing her hands and the bathroom setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the cause-effect as the toothpaste was dispense not accurate and it was not falls even though  it was floating and it was unrealistic as the paste was coming from the toothpaste. [Prompt Alignment]: The subject holds pink tube on her left hand.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/753.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged man with dark brown skin and a muscular build powers through the breaststroke in open water, surrounded by a vast, serene lake under a cloudy sky. The wind causes gentle ripples on the water as his purposeful movements create splashes with every stroke..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Subject movement, interaction and cause-effect are unnatural. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: The man is not doing a breaststroke swimming position..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/754.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young white woman with long, wavy blonde hair stands in a living room with flowers and decorative items in the background. She faces the camera and speaks, occasionally gesturing with her hands. The camera remains steady, capturing her upper body and the cozy room setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Texture issue on the lip and hair / Blurred mouth [Motion & Physical Consistency]: Deformed hand and teeth [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/755.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A white teenage girl creates a bubbly foam in her hands and rubs it onto her face, the suds covering her nose and cheeks as she smiles into the mirror..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage [Motion & Physical Consistency]: The movement of the subject or the hand appears stiff and robotic. The soap seems to be deformed and passes or fuse with her fingers. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: The subject is not white teenage girl. The mirror is place behind her and she did not smile towards the mirror. The movement of the subject or the hand appears stiff and robotic. The soap seems to be deformed and passes or fuse with her fingers. .\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/756.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young South Asian woman stands in a brightly lit living room. She twirls in place with a joyful expression, then jumps up with excitement. The background features a sofa and decorative plants. The camera remains steady, capturing her celebratory movements..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Yes [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: Yes - It looks real and not generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/757.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A light-skinned adult man stands beside a horse in a grassy field. He holds a clear plastic bottle filled with water in his right hand and extends it toward the horse’s mouth. The horse lowers its head and drinks from the bottle. The background shows open grass and scattered trees under daylight. The camera remains steady, capturing the interaction between the man and the horse..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: The man is not light skinned. The horse did not drink the bottle.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/758.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A South Asian woman stands in a brightly lit bathroom. She holds a bottle of gel in one hand and a loofah in the other. She squeezes gel onto the loofah, then uses her hands to work the gel into a rich lather. The camera remains steady, showing her actions against a tiled shower background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: No [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: No - It looks like it was generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/759.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A teenage South Asian woman with braided hair walks through a busy night market, using sturdy green crutches for support. Neon lights reflect off the wet pavement, and colorful stalls line the background. The camera follows her at a steady pace, capturing her movement and the lively, rain-soaked atmosphere..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: The overall movement is unnatural, and there is a visible distortion in her hands. Stabilize to make it look normal. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/760.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A white towel flattens against the glass door as the laundry machine spins at full speed, the rhythmic motion framed in soft, warm lighting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Cause-effect failure as the towel didn't fall. [Prompt Alignment]: The tower is not flattened against the glass door.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/761.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Latina woman with long dark hair sits at a vanity in a softly lit bedroom. She holds a spoolie brush in her right hand and gently brushes her eyebrows upward, shaping them with smooth strokes. The camera remains steady, focusing on her face and the grooming action..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: She holds a spoolie brush in her left hand.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/762.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: {\n  \"description\": \"A sleek, minimal studio setup with a Tinkies Strawberry Bar placed on a rotating pastel-purple platform. The camera captures tight macro shots of the texture, strawberry filling oozing slightly, and the golden sponge. Background elements include hovering strawberries and soft ripples of color echoing the flavor. No text.\",\n\n  \"style\": \"minimal, modern, sensory-focused\",\n\n  \"camera\": \"macro close-ups rotating around the bar, zooming into texture and filling\",\n\n  \"lighting\": \"controlled studio light with pink backglow and subtle highlights\"\n}.\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Cause-effect fails as the strawberry moves on its own. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/763.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman with curly hair sits on a couch in a cozy living room. She holds a small handful of jelly beans in her right hand, selects one, and pops it into her mouth. She leans back slightly, chewing the jelly bean while looking ahead. The background shows soft lighting and decorative pillows. The camera remains steady, capturing her relaxed posture and the casual setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Yes [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: No - It looks like it was generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/764.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The person stands barefoot on wooden flooring, holding a Shark vacuum cleaner upright. The person subtly shifts weight from one leg to the other, with stairs and a doorway visible in the background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The texture of the hose on the vacuum and some of her tattoos are flickering. [Motion & Physical Consistency]: Good. [Prompt Alignment]: Good.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/765.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A group of children of various races, ages 7 to 10, run together across a wide-open grassy field scattered with wildflowers. One child with freckles holds the string of a green kite, making it fly as it twirls and flips in the bright blue sky above. The camera captures the children’s movement and the colorful field under clear daylight..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the subject interaction as the hand of the subject was seen having a deformation. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/766.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Southeast Asian woman sits at a vanity in a softly lit bedroom. She lifts a bunny-ear headband with both hands and places it onto her forehead, adjusting it to secure it in place. The background shows a mirror, makeup items, and a small lamp on the vanity. The camera remains steady, capturing her focused movements as she gets ready..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: good. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/767.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black man with a fit build stands outdoors in front of a gray house with a window on the left. He wears a white quilted zip-up jacket, light blue jeans, and a maroon baseball cap. A wedding ring is visible on his left hand. He looks directly at the camera, speaking in a calm, conversational manner while maintaining steady eye contact. Natural daylight brightens the scene, with green grass and trees in the background. The camera remains steady, capturing his upper body and subtle facial expressions..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: The subject is not young. A wedding ring is visible in his right hand.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/768.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young White woman with an athletic build sits confidently in a modern gaming café. She wears a black hoodie with neon accents and straight-leg jeans. The background shows sleek gaming setups and softly lit shelves with gaming memorabilia. She looks directly into the camera, maintaining steady eye contact and a calm, composed expression. Her movements are minimal and natural, conveying confidence and approachability. The camera remains fixed at eye level, capturing her relaxed posture and the ambient lighting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The texture of the mouth is flickering. [Motion & Physical Consistency]: The eye and head movement looks robotic; the deformation of the hand is unnatural. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/769.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A group of three friends—a Southeast Asian woman with long dark hair, a Black man with short hair, and a white teenage boy—sit together at a sunlit café table. Each holds an iced coffee cup. They smile and clink their cups together, then bring the cups to their mouths and take a sip. The background shows large windows with sunlight streaming in and café decor. The camera remains steady, capturing their cheerful interaction..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The wordings are garbage symbols. [Motion & Physical Consistency]: Good. [Prompt Alignment]: They didn't clink their cups.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/770.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A child with curly hair and rosy cheeks joyfully squishes the top piece of pillow-soft sandwich bread onto a lunchbox sandwich filled with colorful vegetable layers, seated beside a window with sunshine illuminating their playful action in a cozy family kitchen..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Good [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/771.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: On a futuristic glowing platform surrounded by floating holographic screens, a man with dark brown skin and tightly coiled hair uses a sleek, chrome megaphone to deliver a message. His gestures are sharp, and the scenes around him pulse with neon light..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the subject interaction as the hand was having a deformation. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/772.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly South Asian man with salt-and-pepper hair and round glasses is dressed in a warm, dark-green knit cardigan over a white collared shirt and beige trousers. He stands on a charming Parisian street, with iconic cafés and cobblestone roads stretching out behind him. His hands slowly rise above his head, forming a heart shape, while he smiles softly at the camera..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Good.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/773.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A realistic view of space shows the sun at the center, appearing large and radiating light against a dark cosmic background. Multiple ice cream-shaped spacecraft approach the sun from different directions. As they move closer, their surfaces visibly melt, revealing their inner layers. Stars and distant planets are visible in the background. The camera remains steady, capturing the slow movement of the melting spacecraft toward the sun..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: It is not an ice cream-shaped spacecraft and it did not visibly melt.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/774.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman stands in a bright kitchen. She places chunks of red meat into a food processor, then presses the button to grind the meat. As the processor runs, the meat turns into ground beef. The camera remains steady, showing her hands and the food processor on the countertop..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Yes [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: Yes - It looks real and not generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/775.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: Slices of bell pepper, red onion, and whole cherry tomatoes are suspended in mid-air above a hot frying pan. The vegetables appear frozen in motion as they fall toward the pan, creating a vibrant burst of color. The background shows a bright kitchen with a stovetop and utensils nearby. The camera remains steady, focusing on the falling vegetables and the pan below..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is a cause effect issue as some  cherry tomatoes disappeared. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/776.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young East Asian woman with straight black hair sits at a minimalist bedroom vanity. She wears a light gray sweater. She holds a small makeup container in her right hand, twists open the lid, then places the container into a holographic pouch next to several cleanser bottles on the vanity. The background features a tidy vanity table with a mirror and soft natural lighting. The camera remains steady, capturing her actions and the organized setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage [Motion & Physical Consistency]: The movement is not smooth and not natural. The hand appears distorted and the index finger seems to pass through the object. The object is moving or rotating inaccurately. [Prompt Alignment]: There are two subjects. She did not twist open the lid.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/777.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman stands at a bathroom vanity under soft lighting. She squeezes a dollop of cream onto her right palm, then uses her right hand to gently rub the cream into the back of her left hand with smooth circular motions. The background shows a mirror and toiletries on the counter. The camera remains steady, capturing her hands and upper body as she applies the cream..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The cream did not spread or mark accurately. [Prompt Alignment]: She did not squeeze cream.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/778.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged light-skinned man with freckles stands upright on a green snowboard at the top of a snow-covered trail. He faces downhill, then begins gliding down the slope in a crouched position, arms extended for balance. As he descends, he shifts his weight and adjusts his arms to maintain stability. Snow sprays from the board with each turn, and his shadow appears on the snow. The background shows a snowy landscape with pine trees and clear skies. The camera follows smoothly from the side, capturing his movements and the dynamic scenery..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the subject interaction as the body was having a deformation. Lastly with the cause-effect as no visible skid marks was seen on the snow upon hovering the ski on the snow. [Prompt Alignment]: The subject is not clearly visible.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/779.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Black man in athletic clothing holds a plank position on a gym mat, keeping his body straight and parallel to the floor. His forearms and toes support his weight while he looks down. The background shows gym equipment and bright lighting. The camera remains steady, capturing his posture and form..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: No [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: No - It looks like it was generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/780.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A group of adults play kickball on a dirt field bordered by grass and nearby buildings. Several players stand on the field as a woman in a yellow and green jersey prepares to pitch a red kickball. She rolls the ball toward home plate, where a referee stands behind the batter. As the ball reaches home plate, a player kicks it, and multiple players—including one in a yellow jersey and another in black—run toward first base. The referee watches the play, while a player in a yellow and green jersey retrieves the ball near first base. The camera captures the action from a wide angle, showing the field and surrounding area..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Their movements appear stiff and robotic, with noticeable distortions in their hands, legs, feet, and faces. Additionally, there’s a cause-effect issue—when the ball is hit, it suddenly disappears. Stabilize to make it normal. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/781.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman with braided hair stands at a glassblowing workstation in a well-lit studio. She holds a glassblowing pipe in her right hand and carefully blows into the opening, shaping a piece of molten glass. The background shows shelves with glass tools and finished pieces. The camera remains steady, focusing on her hands and the glass as she works..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Her overall movements aren't natural. Interaction with the molten glass where she dips her lips is unrealistic as well as the effects of it are unrealistic. [Prompt Alignment]: She isn't holding a glassblowing pipe and she didn't blow the opening.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/782.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Southeast Asian woman presses a pump nozzle, releasing a swirl of soft foam onto her open palm as the camera captures the satisfying texture buildup..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Cause-Effect fails: Unnatural and Unrealistic application of the lotion. Make it natural. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/783.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Caucasian woman plays a hand clapping game with a group, their hands moving in time with the beat as they smile and giggle..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is deformity in the hands. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/784.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A South Asian woman in traditional attire performs belly dancing in a decorated indoor setting. She moves her hips rhythmically while her scarf flutters with each motion. The background features ornate fabrics and warm lighting. The camera remains steady, capturing her graceful movements and the vibrant setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Yes [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: No - It looks like it was generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/785.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A white teenage girl with light brown hair and mild pore issues stands in a well-lit bathroom. She holds a serum dropper above her forehead and gently squeezes to dispense a single drop onto the center of her T-zone, allowing the serum to touch her skin in a smooth, controlled motion. The camera remains steady, focusing on her face and the precise application..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The amount being dispensed is not accurate and inconsistent. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/786.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young white woman with freckles stands in front of a vanity under soft, even lighting. She uses a dropper to apply small amounts of light foundation to her face. Holding a soft makeup brush in her right hand, she blends the foundation in gentle circular motions across her cheeks and forehead. After blending, she tilts her head slightly and turns toward the camera to display her natural glow. The camera remains steady, capturing her actions and the softly lit background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The cause-effect of the application or blending is unnatural. [Prompt Alignment]: There is no dropper and she did not  uses a dropper to apply small amounts of light foundation to her face.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/787.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Caucasian man wearing a green apron stands in a busy kitchen. He uses a large spoon to scoop cooked shredded chicken from a tray and places it into a mixing bowl. The camera focuses closely on the chicken as it falls into the bowl, highlighting its texture and the juices dripping onto the meat. The background shows kitchen counters and utensils under bright lighting. The camera remains steady, capturing the main action and the kitchen setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Yes [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: Yes - It looks real and not generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/788.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman with curly hair sits in a sunlit park, her sketchbook resting on her lap as she meticulously sketches a sleek, modern watch. The background is alive with soft sounds of rustling leaves and distant laughter, but her focus remains entirely on capturing the watch's intricate details with graceful hand movements..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: There is a cause-and-effect failure because the sketching in the sketchbook does not accurately reflect how the subject does it. [Prompt Alignment]: The subject is not visible to identify that they are a black woman with curly hair.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/789.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The subject, a South Asian man, squeezes a dollop of lotion into his hand, quickly applying it to his skin to keep it moisturized..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: The cause and effect fail due to the sudden disappearance of the lotion bottle. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/790.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young white boy wearing a bright orange raincoat walks through shallow puddles on a city sidewalk at night. He holds a phone in his right hand, using selfie mode to capture his face as he grins and laughs into the camera with each step. Reflections of city lights shimmer on the wet pavement behind him. The camera remains steady, focusing on his cheerful expression and the lively urban background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: He's not holding the phone in his right hand.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/791.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Caucasian woman sits at a grand piano in a well-lit living room. She plays an upbeat melody, her hands moving rhythmically across the keys. The camera remains steady, capturing her focused expression and the polished piano in the background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: There is deformity in the subject's hands and to the piano tiles. Stabilize to make it look normal. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/792.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A white teenage girl uses a foaming pump to release product into her hand, spreads it across her face with upward motions, rinses with warm water, and touches her cheeks to check the softness..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The cause-effect of the product and the deformation of the hand are all unnatural. [Prompt Alignment]: The subject is not white teenage girl, no water visible and she did not spreads it across her face with upward motions, rinses with warm water, and touches her cheeks to check the softness.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: No. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/793.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A teenage Black woman with her hair tied in two buns wears denim overalls and stands in a sunlit barn. She uses a small hammer to shape a horseshoe on a workbench. In the background, a horse looks over the stall door. The camera remains steady, capturing her focused movements and the rustic barn setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: No [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: No - It looks like it was generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/794.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Black man in his 30s stands in a bright kitchen. He adds fresh spinach leaves, banana slices, and a spoonful of peanut butter into a blender jar. The countertop holds a few kitchen items, and sunlight streams through a nearby window. The camera remains steady, capturing his hands as he combines the ingredients..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Yes [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: No - It looks like it was generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/795.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Middle Eastern woman with long dark hair holds a folded sweater close to her face and gently sniffs it. She smiles softly, appearing pleased by the fresh scent. The background shows a cozy living room with a sofa and soft lighting. The camera remains steady, focusing on her expression and the sweater..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/796.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A woman, wearing a brown leather jacket, is sitting in a car and appears to be crying. She covers her face with her hands, showing distress. The text on the screen reveals her story, stating that she thought her boyfriend loved her until she searched his name on Nova AI and found out he was dating her best friend. She then covers her face with her hands, indicating her emotional state. The text continues to reveal that she was about to drive home because she searched her husband's name on Nova AI and found out he was dating her best friend. She continues to cover her face, showing her ongoing distress..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: There is no text on the screen revealing her story visible.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/797.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: On a warm sunny day at a park, a Caucasian man with short hair drinks water from a glass placed next to a water cooler. The camera focuses on the twisting motion of his hand as he opens the cap, showing the clear water flowing into his glass as he enjoys the cool beverage..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the subject movement as the hand was having stiffness. Also with the object object as the lid of the cooler was having a deformation. Lastly with the cause-effect as the glass of water was not drinks and it was seen floating. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/798.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Middle Eastern woman with long dark hair stands in a well-lit bathroom. She holds a sleek bottle in her right hand and tilts it downward, pressing the nozzle to dispense a glossy cream onto her left palm. The background shows a countertop with toiletries and a mirror. The camera remains steady, focusing on her hands and the flowing cream..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Texture [Motion & Physical Consistency]: Good [Prompt Alignment]: She hold the bottle in her left hand instead of right. The cream was on the back of her hand not her palm on the left.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/799.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A teenage boy with freckles and auburn hair sits cross-legged on the floor of a room, pressing a pencil to paper as he skillfully draws the intricate gears and hands of a watch design on a sketchpad..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Subject movement fails due to the stiffness of the hand movement. Cause-effect fails as the drawing in the sketch pad suddenly disappears.  Stabilize to make it look normal. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/800.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Black woman with blue apron and chef hat adds chopped strawberries, bananas, and spinach into a food processor on the kitchen countertop. The camera zooms in on the fruits as they are added, ready to be blended into a smoothie..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage symbols [Motion & Physical Consistency]: The hand movement is not natural and it goes through the blender. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: The video is unrealistic.\n.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/801.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A white young adult woman in a light blue T-shirt sits at a table in a brightly lit room. She holds a colorful phone package in both hands and tears open the top flap. She removes a white charging cable and a set of colorful stickers from the package, then lifts the phone out with both hands. The camera remains steady, showing her actions and the tabletop..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The object and the wrapper get deformed. There is a small white box that suddenly appears. [Prompt Alignment]: The video does not show her removing the charging cable from the package.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/802.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young man with dark skin and braided hair swims backstroke in an abstract digital ocean filled with neon blue waves. As he moves his arms and legs in a smooth backstroke motion, glowing trails of light appear in the water behind him. The background features stylized, luminous waves and a futuristic digital atmosphere. The camera follows his movement from the side, capturing the fluid motion and vibrant setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: good [Motion & Physical Consistency]: good [Prompt Alignment]: Man is swimming the freestyle instead of the backstroke\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/803.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The video shows a close-up of a white electric floor mop cleaning a wooden floor in a modern living room.First, the mop moves forward across the wooden floor, with the cleaning pads visibly spinning. Then, the mop continues to glide back and forth, covering different sections of the floor. The camera remains focused on the mop head and its movement, with the handle angled diagonally across the frame throughout the video..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: The mopping movement and the cause and effect of the mark from the mop are all seem unnatural. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/804.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A white teenage woman with long brown hair grins as she uses a metal spatula to scoop a portion of bubbling lasagna from a baking pan. Strings of melted mozzarella stretch between the slice and the pan. The background shows a bright kitchen with countertops and utensils. The camera remains steady, focusing on her action and the lasagna..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Cause-effect failure as some part of the lasagna disappears inaccurately, though it was not scooped by the spatula. [Prompt Alignment]: Good.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/805.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: Montrer la robustesse de la tente Quechua durant des conditions climatiques complexes : de la tempête de neige dans la montage puis tempête de sable dans le désert du Sahara puis de la pluie diluvienne en plein milieu d'une forêt. La vidéo peut durer 30 secondes et doit montrer la qualité du produit. Toutes les scènes doivent être respectées. CONDITIONS CLIMATES EXTREMES..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The snow's movement is not accurate and inconsistent. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/806.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Middle Eastern man responds to a sidewalk interview in downtown, gripping the microphone and gesturing animatedly while his group listens and cheers him on..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The movement of the man's mouth while talking appears unnatural as it was distorted. There is also an object interaction issue as the cars on the background was deforming. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/807.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A slim South Asian young woman with long black hair stands in a bright indoor gym, wearing a light green top. She is shown from the waist up against a light-colored background. Facing the camera, she maintains steady eye contact and speaks directly with a friendly, engaging expression. The atmosphere is calm and professional. The camera remains steady, focusing on her as she talks. No other people or objects are visible in the scene..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Good. [Prompt Alignment]: She is not a South Asian woman. She is not shown from the waist up.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/808.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: Inside a futuristic medical lab with glowing neon blue panels, a young boy with curly hair carefully places his arm under a sterile light band. A close-up reveals minor scaly skin patches and fine hairs standing upright due to the bright light. His small hand delicately presses the area, as holographic text reading 'If you have this' scrolls down the textured surface of his skin. The rest of the lab hums softly in the background, adding a high-tech atmosphere..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage symbols, not English. [Motion & Physical Consistency]: There is no blinking of the eyes, which is unnatural. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: No. Subject-Object Interaction: Another hand presses the area, not the boy, which is not stated in the prompt..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/809.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged Black woman with short curly hair refreshes the bedding on a bed. She carefully smooths out a blue comforter with a subtle sheen. The surroundings are simple, with plain walls and a large bedframe, as she arranges the bedding with care, creating a calm and orderly atmosphere..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: His overall movements aren't natural and his hand deforms while interacting with the comforter. [Prompt Alignment]: She doesn't look like a black woman and she didn't smooth out the blue comforter.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/810.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A woman in her 30s with light skin and shoulder-length brown hair wears a loose gray T-shirt and sweatpants. She stands in a bright living room with a wooden floor. Holding a cordless vacuum in both hands, she pushes it smoothly and steadily across the floor, collecting crumbs and dust. The background shows a sofa and a coffee table. The camera remains steady, capturing her cleaning motion and the tidy room..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the subject movement as the feet was having stiffness and not smooth upon moving. Also with the subject interaction as the hand and the feet was seen having deformation. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/811.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A woman in her mid-30s with shoulder-length blonde hair in loose waves stands barefoot on a wooden floor by a mirror in a softly lit bedroom. She wears a light beige knit sweater and white linen trousers. Morning sunlight filters through white curtains, illuminating her with a soft golden glow. Pastel-toned outfits—a light blue shirt, cream blazer, and sage green dress—hang on a clothing rack nearby. She gently smiles, looks at the mirror as if choosing an outfit, adjusts her sweater sleeve, and then faces the camera, speaking warmly into a small beige microphone. The handheld camera drifts subtly with shallow focus, highlighting the peaceful, cozy ambiance and authentic atmosphere..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The texture of the sweater is flickering. [Motion & Physical Consistency]: Deformed hand and the cause-effect of the reflection that is not accurate are all unnatural. [Prompt Alignment]: She did not adjusts her sweater sleeve.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/812.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly South Asian man dressed in a beige kurta sits cross-legged in a quiet garden at sunset. He carefully places a cracked silver smartphone on a flat stone and brings down a heavy metal hammer. The screen bursts apart on impact, shards catching the warm, fading sunlight as they scatter..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: His overall movements are unnatural. Object-Object interaction fails as the hammer fails to hit the phone . It breaks unrealistically. The splashes of broken glass are unrealistic. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Unrealistic effects..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/813.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Southeast Asian woman stands at her vanity in a softly lit bedroom. She leans close to the mirror, holding a clear gel brush in her right hand. She gently combs her eyebrows with the brush, moving it along each brow to shape them. Her left hand rests lightly on the edge of the vanity. The background shows makeup items and a small lamp on the vanity surface. The camera remains steady, focusing on her face and precise hand movements..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The sudden changes of her eyebrow is unnatural. [Prompt Alignment]: It's not a clear gel brush and she's not standing.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/814.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The person subtly shifts her weight from one leg to the other, drawing attention to the blue plaid dress and puffed sleeves as the camera slowly zooms in..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Texture issue on the dress [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/815.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A tall Black man in a sharp suit rides a sleek, modern white unicycle through a bustling urban plaza at twilight, the city lights beginning to glow as people stop and watch in amazement at his effortless skill..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage [Motion & Physical Consistency]: The movement and the background movement is stiff and not smooth. There is a subject in the background wearing white shirt sitting on a bench that appears the body got deformed or passes through. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/816.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Caucasian woman in a work outfit pours a glass of sparkling water from a bottle into a cup. The camera zooms in on the bubbles rising to the top of the cup, adding a refreshing touch to her mid-morning break..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: The effects of sparkling water are unrealistic. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/817.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A well-dressed Middle Eastern man and a well-dressed Middle Eastern woman stand together in a modern living room. The man holds a tablet in his right hand and gestures toward the screen, while the woman nods, showing interest. Several colorful shopping bags are arranged on a table nearby. The camera slowly pans left as the man taps the tablet, displaying special offers and exclusive registration benefits. The woman then picks up a gift card from the table and examines it, highlighting the convenience and personalization of the discounts. The background features contemporary furniture and soft lighting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Texture issue. [Motion & Physical Consistency]: There is distortion in the video. [Prompt Alignment]: The man didn't hold a tablet in his right hand.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/818.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An adult man wearing work clothes stands on a metal ladder in the living room of a house. He holds a screwdriver in his right hand and adjusts a screw on the top of a white air conditioner unit mounted on the wall. The air conditioner is connected to a white hose and a black hose, which extend upward and outward from the unit. The background shows a sofa and a coffee table under bright indoor lighting. The camera remains steady, capturing the technician’s actions and the air conditioner installation..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The texture of the side of the conditioner is flickering. [Motion & Physical Consistency]: The hand deformed. [Prompt Alignment]: We cannot assume if it is a screw and he is not adjusting.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/819.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young North American White man, about 40 to 45 years old, stands in a modern warehouse. He has short, pale blond hair and bright blue eyes. He wears a stylish work vest in black, gray, and red with red zippered pockets and reflective stripes, along with khaki cargo pants. Facing the camera, he speaks while organizing items with his hands, maintaining eye contact throughout as he tidies the objects. Behind him, neatly arranged shelves filled with goods create a bright, orderly background, highlighting an efficient e-commerce workspace. The camera remains steady, focusing on the man and his actions..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: He's not organizing items with his hands.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/820.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman with short curly hair stands in a bright kitchen, wearing a casual T-shirt. She holds a bottle of soy sauce in her right hand and pours it directly over a plate of barbeque-glazed ribs on the counter. The background shows kitchen cabinets and utensils. The camera remains steady, focusing on her action and the food..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage Symbols. [Motion & Physical Consistency]: Cause-effect fails as the soy sauce is not spread accurately. [Prompt Alignment]: Good.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/821.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A sleepy Black woman stands in front of the mirror in her pajamas, holding up two outfits against her body and tilting her head as she decides what to wear..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The woman’s hand movement is stiff. There is a subject-object interaction issue, as the pajama she is holding is deformed. Additionally, there is a cause-and-effect issue, as the other pajamas are floating. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: Three outfits are visible in the video, but she is only holding up one. The woman’s hand movement is stiff. There is a subject-object interaction issue, as the pajama she is holding is deformed. Additionally, there is a cause-and-effect issue, as the other pajamas are floating. I believe the video is AI-generated because of this discrepancy..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/822.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black man in an olive-green t-shirt stacks two large Amazon boxes neatly in a sunlit living room with earth-toned decor..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage text. [Motion & Physical Consistency]: His overall movements are unnatural. Interaction with the boxes are unnatural as well as it also glitches and multiplies unrealistically. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: There are more than two boxes in the video and he didn't stack two of it..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/823.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young woman presses a soft pink toothpaste tube, and a swirl of paste coils neatly onto a toothbrush under warm bathroom lighting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage text. [Motion & Physical Consistency]: Her overall movements are unnatural. Object-Object interaction such as toothpaste and the toothbrush is unnatural as well. Also there is already some toothpaste in the toothbrush which cause-effect fails. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: The toothpaste didn't swirl the paste coils neatly onto the tooth brush,.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/824.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Black man in business casual clothing stands in a modern office with desks and chairs in the background. He pushes an upright vacuum cleaner across the carpeted floor, guiding it back and forth as the vacuum picks up visible dust. The camera remains steady, capturing the man’s movements and the clean workspace..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Yes [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: Yes - It looks real and not generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/825.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A woman in a colorful dress with a black crossbody bag walks forward on a path lined with greenery and rocks. Behind her, a woman in a white shirt and black shorts raises her hand to make a peace sign. Another woman in a blue shirt and pink shorts walks beside her, looking at her phone. The background shows a clear sky with clouds and additional people walking along the path. The camera remains steady, capturing the group’s movements and the natural setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Blur face and legs/Flickering [Motion & Physical Consistency]: Deformed hand and face [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/826.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A South Asian man stands at a kitchen counter under bright lighting. He holds a chef’s knife in his right hand and a sharpening steel in his left. He draws the blade along the sharpening steel in smooth, controlled strokes, causing the metal edge to glint with each motion. The background shows kitchen utensils and a tiled backsplash. The camera remains steady, focusing on his hands and the sharpening process..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: No [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No - The video does contain nudity or harmful content. Cause-Effect: No - It looks like it was generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/827.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Indonesian woman with straight black hair stands against a plain indoor wall. She wears a short-sleeve t-shirt. She raises her right arm to expose her underarm, looking at the camera with a tense expression as if straining. The camera remains steady, clearly framing her upper body and face in well-lit, neutral surroundings..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Texture on the edge of the hair. [Motion & Physical Consistency]: Good [Prompt Alignment]: The woman did not raise her arm to expose her underarm.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/828.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: Beneath the fluorescent lights of a bustling urban gym, a dark-skinned bodybuilder clad in a teal tank top stands beside a black punching bag. As the camera closes in on his exposed elbow, it reveals rough, chalky skin contrasting sharply against the sleek gym background. With a slow, deliberate motion, he brushes his other hand across the area, and bold white text overlays the muscle definition, reading 'Even athletes...' The distant background chatter and clanging weights fade into hushed focus, emphasizing the skin moment..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: There is garbage text visible in the video. [Motion & Physical Consistency]: The movement of the man while rubbing his arm is not natural. There is also a cause-effect issue as the chalk texture appeared on areas he did not touch. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: The overlay text is not 'Even athletes...' The movement of the man while rubbing his arm is not natural. There is also a cause-effect issue as the chalk texture appeared on areas he did not touch. I believe the video is AI-generated because of this discrepancy..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/829.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged South Asian woman wearing a light gray jumpsuit jogs steadily up a rugged mountain trail. Her water bottle is clipped to her side as she maintains a confident pace, arms swinging naturally. The background shows rocky terrain and sparse vegetation under a bright midday sun. The camera follows her from the side, capturing her movement and the scenic landscape..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: The movement appeared stiff and not smooth. Her legs deformed. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/830.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Southeast Asian woman with long black hair, wearing a light-colored T-shirt, stands at a bathroom sink. She presses the nozzle of a cleanser bottle to dispense product onto her palm, then rubs the cleanser over her face in smooth circular motions with both hands. The background shows a bathroom mirror and tiled wall. The camera remains steady, focusing on her cleansing routine..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is a deformity in the subject's hands. Cause-effect fails as there is a liquid that comes out even though it was not pressed by a  hands. [Prompt Alignment]: The subject did not press the nozzle.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/831.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged man wearing a dark suit stands in a parking garage at night in front of a black car with a modern design. He gestures with his hands while speaking, occasionally turning his head to look at the car and then back at the camera. The background features a multi-level parking structure with illuminated buildings and parked vehicles. The camera remains steady, capturing the man’s upper body and the surrounding area..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The deformed hand and the cause and effect of the reflection seem unnatural. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/832.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman with curly hair is standing at a rustic wooden kitchen counter under soft morning light, carefully rolling out golden pastry dough with a wooden rolling pin. Flour dusts her apron and the countertop, while a bowl of mixed ingredients sits nearby, suggesting preparations for a homemade pie..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There are no marks or changes on the dough. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/833.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The person gently shifts weight from one leg to the other, subtly adjusting the jacket while holding the phone steady for a clear mirror selfie..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Her left hand is deformed. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/834.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A close-up of a hand lifting a pepperoni pizza slice from the pie, with golden cheese stretching in long, gooey strands under warm kitchen lighting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The movement when lifting the pizza, the spatula and the pizza is unnatural as well. The pizza is boiling unrealistically. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: No. Object Existence: No. Subject-Object Interaction: There is no hand lifting the pizza. The spatula wasn't mentioned either and the pizza isn't lifted by the hand either..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/835.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The camera glides down a long queue of people standing outside a trendy shop—passing a Black teen in headphones, a white woman in a trench coat sipping coffee, a South Asian man checking his phone, and a Southeast Asian girl chatting with a friend..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: The hand of the second man deformed when he swung it and some faces of the other subject deformed as well. Stabilize to make it look normal. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/836.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black man in a casual shirt stands at a kitchen counter with a steaming plate of spaghetti. He holds a jar in his right hand and pours marinara sauce over the pasta, letting it cascade evenly across the noodles. He then uses a fork to stir the sauce into the spaghetti. The camera zooms in on the plate, focusing on the sauce as it coats the pasta..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Yes [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: Yes - It looks real and not generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/837.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A group of Black college students stands together on a busy city street. One man in the group holds a microphone and answers confidently, speaking toward the camera. The other students stand nearby, laughing and reacting to his response. The background shows passing pedestrians and storefronts under bright daylight. The camera remains steady, capturing the group’s lively interaction and the bustling street scene..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The movement is not smooth and not natural. There are deformities and distortions in the background subjects. [Prompt Alignment]: The man is not holding the microphone.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/838.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Latina woman with straight dark hair sits in the driver’s seat of a car, wearing a white shirt and a necklace. The car’s interior, including seats and a window with a building visible outside, is shown in natural daylight. She looks directly at the camera and speaks, her facial expression shifting to quiet astonishment. She maintains steady eye contact and uses minimal hand gestures while talking. The camera remains steady, capturing her sincere demeanor and the casual atmosphere inside the car..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/839.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: On a surreal floating island surrounded by cascading waterfalls, a young Black woman with a shaved head calmly peels away layers of translucent, shimmering foil. Beneath lies a folded garment in a radiant shade of gold, emitting a faint glow. She holds it up, its light reflecting off the lush greenery around her, creating a moment of tranquil magic..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The overall movement of the subject, the sudden disappearance of the paper, and the sudden appearance of the folded garments are unnatural. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: She did not peel away layers of translucent, shimmering foil..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/840.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: In a bustling beauty salon, a middle-aged East Asian woman with her hair tied in a neat bun demonstrates waxing techniques on her client's back, the bright workspace filled with neatly arranged jars of wax..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: NULL [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/841.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A confident woman with light skin and long wavy dark hair stands in a modern, softly lit living room. She wears a black silk and lace corset over a fitted sheer blouse, paired with tailored high-waisted trousers, reflecting a gothic-inspired luxury lingerie outerwear look. She places her hands on her hips, then slowly turns to face the camera with a poised stance, exuding empowerment and inclusive fashion. The camera stays steady, capturing her elegant posture and the sophisticated room decor..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The texture of the corset is flickering. [Motion & Physical Consistency]: The hand deformed. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/842.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A muscular man with dark brown skin stands on an urban rooftop, wearing a white tank top and blue shorts. He catches a slightly worn softball with his bare hands, gripping it firmly. The background shows gritty concrete and a distant city skyline. The camera remains steady, capturing the man’s action and the rooftop setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage [Motion & Physical Consistency]: Good [Prompt Alignment]: The video does not show he caught the ball.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/843.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A teen throws her tablet and charger onto the edge of the bed, where they land beside a hoodie and an open backpack..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The hand movement is not normal and the other bag suddenly appear. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: The video is unrealistic.\n.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/844.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Hispanic man in his 30s with short black hair sits at a kitchen table. He holds a sliced bagel in his left hand and uses a butter knife in his right hand to spread butter evenly across the surface of the bagel. The background shows a bright kitchen with a window and countertop appliances. The camera remains steady, focusing on his hands as he spreads the butter..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The effects of butter when spread aren't consistent and realistic. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/845.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A medium shot shows an adult woman in her 30s with medium skin tone and shoulder-length dark hair, wearing a cream-colored t-shirt and blue jeans. She stands in front of an apartment building entrance during daylight, holding a black wheeled suitcase in her right hand. She wears sunglasses and holds a smartphone in her left hand, recording herself while smiling. The camera slowly pans right as she steps out of the entrance and walks down the sidewalk, continuing to hold the suitcase and phone..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The texture of the wall is flickering. [Motion & Physical Consistency]: The walking movement and the interaction with the deformed hand are unnatural. [Prompt Alignment]: She's holding a black wheeled suitcase in her left hand.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/846.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Hispanic man in a red jacket turns his wrist to check the time on his smart watch while crossing a desert trail. The camera zooms in on the watch as he tracks his steps, with the expansive desert landscape stretching out in front of him..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: There is garbage text visible in the watch. [Motion & Physical Consistency]: The movement of the man appears stiff. There is also a subject-object interaction issue as there was an object came out on his hand. [Prompt Alignment]: The man is not Hispanic.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/847.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A group of young adults sit at tables inside a brightly lit pizzeria with blue and red decor. One young adult holds a slice of pizza in both hands, brings it to their mouth, and takes a bite while their eyes are closed, appearing to savor the taste. The background shows other diners and colorful decorations. The camera remains steady, capturing the group and the pizzeria setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Texture issue with the pizza [Motion & Physical Consistency]: Deformed hand [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/848.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young man in a casual t-shirt stands outdoors by a fishing pond, holding a small bottle of liquid attractant. Facing the camera in selfie mode, he briefly introduces the product with a cheerful expression. Next, he squeezes three drops of the attractant onto his fishing bait before showing its fragrant aroma. He casts the bait into the water with one hand, then the camera cuts to the float starting to sink and the fishing rod bending sharply as he hooks a fish. The man quickly reels in and lifts a medium-sized fish, displaying it to the camera with the attractant bottle in hand. The background reveals the pond and greenery under bright daylight. A rapid-beat track plays during the catch moment. The camera stays steady, capturing all key actions in sequence..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: textural details in hair [Motion & Physical Consistency]: hands are deformed [Prompt Alignment]: good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/849.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: In a bright gym, a Black man with a well-toned physique scoops protein powder from a tub into his shaker. The camera focuses on the scoop as he pours the powder into the bottle, ready to hydrate after an intense workout..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The overlay wordings are just garbage, not English. [Motion & Physical Consistency]: Subject movement failure and Cause-effect failure. Stabilize to make it look normal. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: The subject did not scoops the protein powder from a tub into his shaker..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/850.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Middle Eastern man in a gray hoodie stands at a kitchen counter. He uses a metal spatula to lift a large square of lasagna from the corner of a baking tray, the golden crust crackling slightly as he serves himself. The background shows a bright kitchen with countertops and utensils. The camera remains steady, focusing on his hands and the lasagna..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Cause-effect fails as there is a part of lasagna moved up together with the slice she is getting and it sits back on the baling try with no traces of it being lifted magically. [Prompt Alignment]: The subject's face is not visible, unable to identify race.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/851.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged Black woman wearing safety goggles holds a circular saw with both hands, sparks flying as she carefully cuts through a thick wooden plank on a workbench in a brightly lit workshop surrounded by scattered tools and wood shavings..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The cause-effect of the circular saw is unnatural. [Prompt Alignment]: She did not holds a circular saw with both hands.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/852.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A South Asian man in a casual T-shirt and jeans carries a box of fresh fruits in both hands as he walks steadily into a bright kitchen. The camera remains steady, showing him handling the box carefully while moving through the kitchen space. The background features countertops and kitchen appliances under natural light..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Deformed hands [Prompt Alignment]: He is not a South Asian man and there is no kitchen appliances visible.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: No. Object Existence: No. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/853.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Singaporean Chinese woman with straight black hair is shown in a vertical close-up against a light mocha brown background. She gently massages her neck with both hands, moving her palms upward along the length of her neck in smooth, continuous strokes. The camera remains steady, focusing on her neck and hands as she performs the massage..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: She's not Singaporean Chinese. She's moving her hands downward.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/854.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman in a bright yellow sweater sits on a park bench under the shade of a large oak tree in the late afternoon. She holds a smartphone in one hand and carefully peels the backing off a clear screen protector with the other. She aligns the screen protector over the phone’s surface and presses it down to apply it smoothly. The background shows green grass and dappled sunlight. The camera remains steady, capturing her focused action and the peaceful park setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The finger appears deformed and the interaction between the finger and the film fails. The film being peeled or the way it moves is not accurate. [Prompt Alignment]: She did not align and press it down.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/855.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The subject, a South Asian woman, spreads vegetable toppings onto a baking tray, ready to roast them for dinner..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: The overall movement is unnatural and the hand goes through the bell pepper.\nThe bell pepper suddenly changes. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/856.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Black man pours cereal into a bowl, the pieces spreading evenly as he fills it, ready to enjoy with milk..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Interaction fails as the cereal passes through the container. Cause-effect failure as the cereals being poured on the bowl is inconsistent with whats on the bowl. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/857.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A teenage boy with curly red hair and freckles stands on a train station platform, wearing glossy black tap shoes. He repeatedly taps his feet in sync with the rhythm of passing trains, creating a sharp, electrifying pattern that echoes through the station. Commuters walk past in the background. The camera remains steady, capturing the boy’s rhythmic movements and the busy platform setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: The subject's face in the video is not visible, so we cannot identify whether he is a teenage boy with curly red hair and freckles.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/858.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A cheerful teenage Black girl with curly hair tied in a bun is standing at an outdoor picnic table under a canopy. She cuts into a steaming sausage on a brightly colored ceramic plate, revealing its juicy interior, as soft afternoon light filters through surrounding trees..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Cause-effect failure as the sliced sausage moves on its own. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/859.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged woman with medium brown skin and short black hair glides confidently across the surface of a man-made wave pool on a sleek, metallic windsurfing board. She stands upright, holding onto a neon-lit sail that glows purple and green. The background features a futuristic cityscape with towering skyscrapers illuminated by bright, colorful lights. The camera follows her smooth movement across the water, highlighting the blend of modern technology and sport..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Good [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/860.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Hispanic man in his thirties with short dark hair stands in a kitchen. He holds an open can of wet dog food in his left hand and uses a metal spoon in his right hand to scoop the food into a dog bowl on the counter. The background shows kitchen cabinets and a window with daylight. The camera remains steady, focusing on his hands as he fills the bowl..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The mark is inconsistent and not accurate. [Prompt Alignment]: He did not scoop the dog food.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/861.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Indigenous woman with copper-toned skin sits by a sunlit window. She holds a small serum bottle in her right hand and places several drops of serum onto her cheeks. She sets the bottle aside, then uses both palms to gently blend the serum into her skin. The background shows soft daylight streaming through the window, with a simple interior visible. The camera remains steady, focusing on her face and hands as she completes the skincare routine..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Cause-effect failure as the serum didn't spread and dispensed accurately. [Prompt Alignment]: She didn't hold a small serum bottle in her right hand.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/862.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: Inside a spacious bedroom adorned with elegant furniture and soft lighting, an elderly Black man wearing a tailored gray coat and a matching flat cap strides confidently with a small teal rolling suitcase. The suitcase moves smoothly across the floor, its wheels barely making a sound. He pauses beside a large bed, turning to smile warmly at the camera before gesturing toward the suitcase handle. The camera shifts focus to its durable rubberized wheels and polished corners, contrasting with the serene bedroom ambiance..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Movement of the subject is unnatural. Interaction with luggage as it deforms is not in accordance to physical law. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: I think it's generated by AI\n.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/863.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Middle Eastern man lowers his phone onto a bed after filming, the camera view slowly tilting toward the ceiling as he walks away..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The overlay wordings are just garbage, not English. [Motion & Physical Consistency]: The subject's body movement is not natural and his hand goes through the phone. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: He did not lowers his phone onto the bed and there is no bed visible in the video..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/864.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young East Asian man of average build sits in the driver’s seat of a car, wearing a casual dark sweatshirt and glasses. The car’s interior is visible, with the steering wheel in the foreground. He faces the camera, maintaining steady eye contact as he speaks slowly and clearly. His expression is subdued, with slightly furrowed brows, conveying a quiet sense of concern. The camera remains fixed, capturing his composed demeanor and the car’s interior..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: good [Motion & Physical Consistency]: good [Prompt Alignment]: good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/865.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A fair-skinned young boy wearing a baseball cap sits beneath a sprawling oak tree in a wide, sunny meadow, scratching quick sketches of birds into a leather-bound journal with a sharp graphite pencil clutched in his hand..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: There is stiffness in his hands. Cause-effect failure as the pencil didn't leave consistent marks. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/866.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged White man with a beard kneels on a concrete sidewalk and uses colorful chalk to create a detailed cityscape, his hands smudged with red, blue, and yellow hues as curious onlookers pass by..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: NULL [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/867.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The camera slowly zooms in on the glass of tea, highlighting the tea bag steeping and the surrounding tea bags on the table..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Texture issue on the tea bag / Blurred tea bag [Motion & Physical Consistency]: There is no water movement. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/868.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The person gently holds the handheld device with both hands, subtly shifting his posture to highlight the device’s sleek design..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/869.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman sprays hair spray onto her styled hair, holding the can at arm’s length as she ensures every strand is set..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The video has a garbage caption. [Motion & Physical Consistency]: There is an issue with the subject movement as it appears unnatural. Also with the cause-effect as no visible marks was seen as she sprays into her hair. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: There is an issue with the subject movement as it appears unnatural. Also with the cause-effect as no visible marks was seen as she sprays into her hair..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/870.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: At a busy beach picnic, a young Hispanic man opens a can of tuna to add to his salad. The camera zooms in on his hand as he opens the can, showing the rich, flaky tuna inside as it's ready to be enjoyed with a side of vegetables..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The wordings are garbage symbols. [Motion & Physical Consistency]: There is stiffness in the hands, making it unnatural. Cause-effect failure as the can is already open. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: I think it's generated by AI\n.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/871.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young adult stands outdoors, holding a silver can in their right hand. They raise the can and throw it downward onto a concrete surface. The can lands, and a blue liquid pours out in a steady stream from its base. The camera then cuts to a waterfall scene, where water cascades over rocks into a river below..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The object-object interaction fails and the cause-effect of the blue liquid is unnatural. [Prompt Alignment]: The subject's face is not visible and they did not raise the can and throw it downward onto a concrete surface.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/872.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young woman with light skin and short black hair wears pastel-colored gym attire. She sits on a bench in a well-lit gym, gripping a barbell with both hands. She lowers the barbell toward her chest, then presses it upward in a controlled motion. Several pink shopping bags are arranged on the floor beside the bench. The camera remains steady, focusing on her upper body and the gym environment..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage Symbols. [Motion & Physical Consistency]: Her hand is distorted. [Prompt Alignment]: Good.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/873.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Caucasian woman with shoulder-length straight brown hair sits in a softly lit bedroom. She wears a light gray T-shirt. Holding a small bottle in one hand, she applies oil directly to her scalp, then uses her fingertips to gently massage the oil into her scalp with circular motions. The camera remains steady, focusing on her head and upper body against a background of a neatly made bed and a nightstand..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The effects of oil aren't natural and realistic. [Prompt Alignment]: It wasn't shown that he applied the oil directly onto her hair and she didn't massage the oil in circular motion.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/874.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly Black man with a graying beard sits on a wooden bench in a serene park surrounded by autumn leaves. He carefully unboxes a matte black drone with intricate detailing, holding it up to display its polished surface as sunlight glints off its surface. The camera remains steady, capturing his focused movements and the peaceful park background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The drone propellers deforms unnaturally. [Prompt Alignment]: He isn't an elderly man but more like a middle-aged. There is no box and it wasn't shown he is unboxing either.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: No. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/875.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: Two Black teenagers stand side by side outdoors on a warm afternoon, each holding a smoothie bottle. They bump their bottles together in a playful cheers, then grin and take a long sip at the same time. The background shows green trees and soft sunlight. The camera remains steady, capturing their cheerful expressions and relaxed posture..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/876.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: Two climbers, a man and a woman in their thirties, scale a wet, vertical rock face beside a rushing waterfall. Both wear helmets, harnesses, and waterproof jackets. Each grips a rope with one hand while reaching for holds on the slick rock with the other, carefully moving upward. Water sprays nearby, making the surface glisten. The background shows the waterfall cascading down and mist rising around them. The camera remains steady, capturing their coordinated ascent and the dramatic setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: There is stiffness with their overall movement. Interaction with the subject is not in accordance to physical law as their faces are deformed. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/877.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged man stands in front of a bathroom mirror. He uses his hands to attach snap-on teeth to his existing, decayed teeth, then smiles widely to show a full, even white smile. The background includes a sink and tiled walls. The camera remains steady, capturing his actions and reflection..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Deformed hand and teeth [Prompt Alignment]: There is no bathroom mirror and he did not uses his hands to attach snap-on teeth to his existing, decayed teeth, then smiles widely to show a full, even white smile.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/878.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A close-up shot shows an adult man’s hand holding an iPhone 15 Pro upright against a plain, light-colored background. The man wears a gray sweater. The phone screen displays a list of credit card offers and interest rates. The camera remains steady, focusing on the phone and the man’s hand..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage symbols [Motion & Physical Consistency]: Good [Prompt Alignment]: Phone does not look like an Iphone 15 pro\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/879.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Middle Eastern woman with long dark hair sits at a vanity table in a softly lit room. She dispenses a rich, glossy lotion onto her palms and rubs her hands together, spreading the creamy lotion evenly. The camera focuses closely on her hands, highlighting the lotion’s creamy texture and the soft glow it leaves on her skin. The background shows a mirror and a few beauty products. The camera remains steady throughout the action..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: It appears to pass through each other. The lotion did not spread or mark accurately. [Prompt Alignment]: She did not dispense lotion.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/880.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young White woman with straight brown hair stands at a bathroom sink. She presses the pump of a foaming cleanser bottle with one hand, dispensing product into her palm. She spreads the foam evenly across her face using both hands in gentle upward motions. The background shows a well-lit bathroom with a mirror and tiled walls. The camera remains steady, focusing on her face and hands during the cleansing routine..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The video has a garbage caption. [Motion & Physical Consistency]: There is an issue with the object object interaction failure as the liquid product on the bottle drips on the wrong hole making it appear unrealistic. Also with the cause-effect as the liquid was still dripping even if there is no hole on and the product was inconsistent being applies on the face. [Prompt Alignment]: The subject does not  presses the pump of a foaming cleanser bottle with one hand, dispensing product into her palm.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/881.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly South Asian woman with silver-gray hair sits in a wooden canoe on a narrow river bordered by lush greenery. She gently paddles the canoe forward, then pauses to look up at the sunlight filtering through the trees. The camera remains steady, capturing her relaxed posture and the peaceful riverside scenery..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The paddle glitches and deformed. [Prompt Alignment]: Her face isn't visible to determine if she is South Asian.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/882.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The subject, a Hispanic woman, skips rope with precision, the rope moving in smooth circles as she keeps up the rhythm..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: NULL [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/883.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A muscular Latino man with a strong build and short dark hair stands outdoors on a city street. He wears a fitted gray T-shirt and dark jeans. Buildings, trees, and parked cars are visible in the background under natural daylight. He faces the camera directly, maintaining steady eye contact. His facial expression shifts naturally to show gentle surprise as he speaks. The camera remains fixed, capturing his upper body and the casual urban setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/884.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Middle Eastern woman with short hair walks through a bustling street market in a white blouse and a black skirt. The camera captures her from the side, showing the fluidity of her steps and the lightness of the skirt as she moves through the vibrant scene..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the subject movement as the subject in the background  was having stiffness and not smooth upon moving. Also with the subject interaction as the face and hand on the subject at the background was having a deformation. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/885.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged Caucasian man stands in a well-lit living room. He holds a wireless earbud in his right hand and inserts it into his right ear, gently pressing it to secure the fit. The camera remains steady, focusing on his upper body and the earbud as he completes the action..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: He's not holding the earbud in his right hand.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/886.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Southeast Asian woman with a soft smile leans forward and sets her camera down beside her skincare products, the lens tilting slightly as it rests..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage text. [Motion & Physical Consistency]: Her overall movements are unnatural. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: She didn't set the camera down, it was already down..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/887.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A macro shot of swirled chocolate and vanilla ice cream with glossy fudge ripples and tiny air pockets in the surface..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The movement of the  ice cream are unnatural, it fails and do not happen in accordance to physical laws. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: I think it's generated by AI because of the unnatural movement..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/888.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: In the base of a waterfall, a Black woman with long hair pours lavender-scented body wash into one hand. The camera focuses on her hand as the purple liquid pours smoothly, showing the rich lather forming as she prepares to cleanse her skin..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The effects of body wash when poured are inconsistent and unnatural. [Prompt Alignment]: Her face isn't visible to determine if she is a black woman and she isn't the one who poured the body wash but another person instead.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/889.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Middle Eastern woman with a dark ponytail stands at a bathroom sink under bright lighting. She dispenses gel cleanser from a pump bottle onto her fingertips, then gently rubs the cleanser over her T-zone. She splashes water across her face to rinse. The camera remains steady, showing the sink and bathroom counter in the background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage Symbols. [Motion & Physical Consistency]: The product being spread to her face was inaccurate. [Prompt Alignment]: She didn't rub cleanser over her T-Zone.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/890.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged Mexican man with a mustache opens a packet of oatmeal cookies in his kitchen. The camera shows his hand as he tears the packet, revealing the soft, homemade-style cookies inside, perfect for a quick snack..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The wordings are garbage symbols. [Motion & Physical Consistency]: Interaction with the packet as it deforms is not in accordance to physical law. [Prompt Alignment]: He didn't tear the packet.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/891.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly Black man with a salt-and-pepper beard sits at a wooden table outdoors on a sunny afternoon. He wears a light button-up shirt and glasses. Holding a compact wood-burning tool in his right hand, he carefully carves the outline of a bird into a slice of unfinished wood held steady with his left hand. The table is scattered with small woodworking tools and wood shavings. The camera remains steady, capturing his focused movements and the bright, relaxed outdoor setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Cause-effect fails as the marks created are inaccurate. [Prompt Alignment]: Good.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/892.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Latina woman maneuvers a pastel pink spatula with precision. She spreads a milky cream across a soft surface that subtly shifts hues under the cream's weight. The camera remains top-down, focusing on her smooth, controlled strokes as the spatula leaves behind silky trails in the material. Light reflects off the glossy cream, creating a soft, glowing ambiance..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Cause-effect failure when she flipped the object she was holding, it was too clean. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/893.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young East Asian man sits indoors in front of a vanity shelf. He holds a colorful bag of Jelly Belly jelly beans in his right hand, resting it on his lap. He glances at the bag, then looks up at the camera, adjusting his posture slightly. The background shows shelves with beauty products and soft indoor lighting. The camera remains steady, capturing his relaxed pose and subtle movements..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Good [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/894.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young White woman stands confidently in a sleek urban financial office, shown from the waist up in a medium shot. She wears a tailored navy blazer over a crisp white blouse with a mid-coverage neckline and covered shoulders. Her hair is neatly styled, and she wears minimal, elegant accessories. She faces the camera directly, maintaining steady eye contact as she speaks with clarity and assurance, keeping a calm, steady posture without using hand gestures. The background is clean and uncluttered, featuring a dark wooden desk with rounded edges, a high-back leather office chair, a ceramic coffee mug, a closed laptop, and a leafy green plant near a window. Soft natural light fills the scene, and the camera remains steady at her eye level..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/895.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: In the heart of the city, a South Asian woman with long straight hair walks down a busy Tokyo street in a light pink shirt and a matching skirt. The camera zooms in on her lower body, showing the soft flow of the skirt as she walks confidently..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage symbols, not English. [Motion & Physical Consistency]: The overall walking and her eyes movement of the subject is unnatural. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: The subject in the video is not a South Asian woman..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/896.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly East Asian man stands on the balcony of a high-rise apartment at night. The skyline is bathed in neon hues, and he gazes at a smartphone with shimmering futuristic apps. The glowing interface mirrors brilliantly in his glasses as he scrolls with his weathered fingers..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Cause-effect failure on the screen as it is not accurate. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: I think it's generated by AI\n.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/897.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: Several bright pink juicy clear boba spin together with whole strawberries, blackberries, raspberries, and blueberries. The berries move in slow motion and boba rotate smoothly as a loose cluster, swirling The camera stays steady, clearly capturing the ongoing rotation and vibrant colors as the fruit and boba move in unison..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There are boba and fruits that are merging. They are moving on their own and unnaturally. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/898.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Caucasian woman takes a long sip from her wine glass, letting the deep red liquid settle on her tongue..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: The subject's movement is weird and unnatural. Subject–object interaction failure due to the subject's lips passing through the glass of wine. Stabilize to make it look normal. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/899.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A woman with braided hair wearing a colorful patterned dress sits upright on the back of a gigantic elephant as it slowly walks along a calm beach at sunrise. Gentle waves brush against the elephant’s feet, reflecting pink and orange hues from the sky. The background shows the sandy shoreline and the soft glow of early morning light. The camera remains steady, capturing the woman’s relaxed posture and the peaceful beach setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Flickering issue on the elephant's ear. [Motion & Physical Consistency]: Deformed foot and the cause-effect of the elephant's footprint is unnatural. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/900.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged South Asian man with salt-and-pepper hair stands at a worktable under the soft glow of fairy lights at midnight. He holds a textured green book cover with both hands and carefully presses it into a metal embossing machine. As he rolls the lever, gold foil on the cover shifts and reveals detailed lettering on the spine. The camera remains steady, focusing on his hands and the embossing process against the illuminated background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The overall movement is not fluid and unnatural. [Prompt Alignment]: He did not rolls the lever, gold foil on the cover shifts and reveals detailed lettering on the spine.\nThere is no Gold foil.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/901.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The camera slowly zooms in on the polished black leather shoes, highlighting their sleek design and textured sole against the softly blurred background of hanging clothes..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/902.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A person kneels beside their car, using a compact vacuum to remove dirt and debris from the floor mats and seat crevices..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: The subject's movement is stiff and looks robotic as the subject's movement is stiff, not smooth. There is deformity in the subject-subject interaction in the left hand. Stabilize to make it look normal. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/903.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Caucasian woman with short hair uses a makeup brush to apply contour to her face. The camera focuses on her hand as she gently blends the product into her skin, sculpting her jawline with precision and a natural-looking finish..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The movement of the woman while applying the contour is not natural as it appears stiff. There is also a cause effect issue as it was not applied evenly. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: The subject can't be determined due to lack of view. The woman did not apply the product on her jawline. The movement of the woman while applying the contour is not natural as it appears stiff. There is also a cause effect issue as it was not applied evenly. I believe the video is AI-generated because of this discrepancy..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/904.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: In a spacious park, a young man with a beard wearing a maroon suit fills a glass with water from a stainless steel water pitcher. The camera focuses on his hand as he pours, showing the crisp, clean water flowing into the glass, ready to enjoy under the trees..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Blurry foreground [Motion & Physical Consistency]: The facial gesture and the water drips unnaturally. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: There is another person in the video..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/905.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman with curly hair, wearing a light floral summer dress, stands at a picnic table outdoors. She holds a clear jug filled with vibrant red fruit punch in both hands and steadily pours the punch into a glass cup. Slices of fruit float in the punch, catching the sunlight as the liquid flows into the cup. The camera remains steady, focusing on the pouring action and the colorful drink, with the picnic table and bright outdoor setting in the background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the cause-effect as the red fruit punch was already being fills but it was having a delayed on the volume as the pitcher was already dispensing a huge amount of the the liquid but still the glass does not full right away making it appear inaccurate. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/906.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Middle Eastern man in a white shirt scoops a rich, creamy chocolate chip ice cream from a large tub. The camera captures his motion as he presses the scooper into the ice cream, lifting the perfect scoop and transferring it to a bowl..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Deformation of  the scoop. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/907.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman carefully opens a cardboard box with scissors, cutting through the tape with precision to reveal the contents inside..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Cause and effect fails because the tape moves on its own without the scissors cutting it. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/908.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young white man stands on a basketball court during an arena event, holding a basketball in his right hand. He jumps toward the hoop and dunks the ball with his right hand. After the dunk, he lands on the court and briefly turns to face the crowd, raising his right arm. The camera follows his movement, capturing the dunk and his reaction. The background shows the arena and spectators..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Actions are too stiff and not smooth. interaction failure, deformation of fingers. Cause-effect failure, the ball was gone. Stabilize to make it look natural. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/909.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A white man in his 30s with short brown hair wears a structured navy blazer over a white shirt and dark jeans. He leans casually with one shoulder against a brick wall on a SoHo street. The background shows boutique storefronts and classic New York City fire escapes under bright daylight. The camera remains steady, capturing his relaxed posture and the urban setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: texture issue of beard [Motion & Physical Consistency]: good [Prompt Alignment]: good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/910.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A white man in a hoodie applies a layer of cleaning foam to a stained countertop, the foam spreading across coffee rings and crumbs..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Movement is stiff and not smooth. The bubbles came out from hand is too much when compared to an ordinary person can hold. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/911.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The camera glides down a long queue of people standing outside a trendy shop—passing a Black teen in headphones, a white woman in a trench coat sipping coffee, a South Asian man checking his phone, and a Southeast Asian girl chatting with a friend..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The movement of the people in the video is not natural as their walk and faces are distorted. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: The white woman in a trench coat sipping coffee. The movement of the people in the video is not natural as their walk and faces are distorted.  I believe the video is AI-generated because of this discrepancy..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/912.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black man with a muscular build and short, neatly styled hair stands in a well-lit gym locker room with white walls and lockers in the background. He wears a fitted white short-sleeve shirt. Facing the camera, he maintains steady eye contact and speaks in a calm, casual manner. His facial expressions briefly shift to show mild dissatisfaction. The camera remains fixed, capturing his relaxed posture and the bright, tidy setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Texture issue on the hair, skin and lip [Motion & Physical Consistency]: Deformed teeth [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/913.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A white teenage girl stands at a bathroom sink under bright lighting. She holds a bottle of foaming cleanser in her right hand and presses the pump, dispensing a dollop of foam onto her left palm. The foam expands slightly as it settles on her skin. The background shows a mirror, tiled walls, and bathroom toiletries. The camera remains steady, focusing on her hands and the product..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The cause-effect of the cleanser is unnatural and the overall movement is not fluid. [Prompt Alignment]: The subject's face is not visible.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/914.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A white teenage girl with long, damp hair lathers shampoo at the crown of her head, her arms raised as suds spread through her strands in a soft backlit shower..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: The subject's movement is stiff and looks robotic as the subject's movement is stiff, not smooth. Subject-Subject interaction fails because there is deformity in the subject-subject interaction while she mixes the shampoo in her hand. Stabilize to make it look natural. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/915.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged traveler adds seasoning powder directly to spiced meat in a cozy kitchen..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The wordings are garbage symbols. [Motion & Physical Consistency]: Good. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: I think it's generated by AI\n.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/916.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly South Asian man with gray hair and glasses stands in a lantern-lit kitchen. He holds a bottle of oil in both hands, twists open the cap, and pours the oil directly onto braised lamb on a serving dish. The background shows warm lighting and rustic kitchen decor. The camera remains steady, capturing his actions and the cozy atmosphere..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Cause-effect failure as oil flows late and inaccurate. [Prompt Alignment]: He's not South Asian. He didn't hold the bottle with both hands.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/917.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young woman with long, iridescent hair and a shimmering mermaid tail moves gracefully underwater inside glowing coral caves. She holds an enormous glass needle in one hand and threads it with shimmering, seaweed-like strands using the other. Her flowing movements create small trails of bubbles. Sunlight filters through the water, casting shifting, kaleidoscopic patterns on the coral walls. The camera remains steady, capturing her focused threading and the vibrant, surreal underwater setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: There is deformity in the hand. Cause-effect fails as the strands become long and some parts of it disappear. [Prompt Alignment]: No needle is visible. She didn't hold an enormous glass needle in one hand and threads it with shimmering, seaweed-like strands.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/918.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Black man uses a traditional razor to shave his face, paying close attention to his facial contours for a clean shave..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Cause-effect failure as the marks of the razor are not accurate. [Prompt Alignment]: Good.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/919.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A teenage girl with straight brown hair, wearing jeans and a pink t-shirt, walks down the brightly lit aisles of a TJ Maxx store. She glances at clothing racks and shelves filled with handbags and accessories, occasionally brushing her hair back with one hand. The camera follows her as she slows to examine a display, then continues walking into a new aisle while looking at the merchandise. The background shows neatly arranged products and overhead lighting. The camera smoothly pans right to capture her movement..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage symbols [Motion & Physical Consistency]: The body movement and the deformed hands are unnatural. [Prompt Alignment]: It is not a TJ Maxx store and she is not brushing her hair with one hand.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: No. Object Existence: No. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/920.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Caucasian woman in a fitted ballet costume performs a series of light leaps across a polished stage, her ballet slippers gliding smoothly with each landing. The background shows stage curtains and soft lighting. The camera remains steady, capturing her movements and the elegant setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: The subject's fingers and legs are deformed and the movement is unnatural. Stabilize to make it look natural. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/921.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An overhead shot of creamy product being spread in looping motions, forming swirls and peaks on a colored acrylic sheet..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is a cause effect issue as the object is empty but dispensed a cream continuously. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: There is a finger visible in the video but was not mentioned in the prompt. The cream was not spread  in looping motions. There is a cause effect issue as the object is empty but dispensed a cream continuously.  I believe the video is AI-generated because of this discrepancy. .\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/922.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young East Asian man wearing a wetsuit rides a red surfboard on a powerful ocean wave. As he approaches the crest, he launches off the board, performing an aerial trick above the wave before landing back on the water. The background shows vivid blue ocean with white frothy waves and bright sunlight reflecting off the surface. The camera follows his movement smoothly, capturing the dynamic action and the energetic scene..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: No [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No - The video does contain nudity or harmful content. Cause-Effect: No - It looks like it was generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/923.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black man in a modern shower pours soothing aloe body wash into his hand. The camera zooms in on his hand as he dispenses the light green liquid, capturing the smooth, refreshing texture as it's ready to be used on his skin..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage symbols, not English. [Motion & Physical Consistency]: The hand movement, the deformation, and the cause-and-effect of application all seem unnatural. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/924.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A group of friends, diverse in age and ethnicity, gather in a neon-lit room with glowing hoops in their hands. One middle-aged woman with auburn hair gracefully twirls her fluorescent purple hoop, her movements synchronized with the pulsing music..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: movement of the hoops does not look natural.  hoops went through the man's arm at the back. Cause-effect failure for the hoop, it became 2. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/925.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young East Asian man with short hair wears sunglasses that reflect the lights of a bustling nighttime market. Standing in the crowd, he tilts his head slightly, his steady pose exuding a composed vibe as pinks and blues from the surrounding lights illuminate the scene..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: The subject can't be determined as he is wearing sunglasses. .\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/926.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Southeast Asian teenage girl with straight black hair is shown in a close-up shot at a vanity. She holds a mascara wand in her right hand and gently wiggles the brush at the base of her upper lashes, then sweeps it upward to separate and lengthen her lashes. The background is softly lit, with makeup items visible on the vanity. The camera remains steady, focusing on her eye and the mascara application..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The mascara reached the lower part of her eyebrow, but no mark was left, which is unnatural. [Prompt Alignment]: It is not visible if it is a vanity and she holds a mascara wand in her left hand.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/927.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly South Asian man in a crisp white linen shirt and tan trousers walks out of a traditional open-air marketplace at golden hour. He carries several colorful woven shopping totes on his arms, with handcrafted jewelry visible inside the bags. The background shows vibrant fabric stalls and vendors bathed in warm sunlight, with long shadows cast across the market entrance. The camera remains steady, capturing his stride and the lively marketplace atmosphere as he moves toward the exit..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: He did not carries shopping totes on his arms.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/928.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A man with straight black hair stands in front of a mirrored wall. He wears a black leather jacket and loose silver pants. Holding a smartphone in his right hand, he flexes his left hand and adjusts the jacket while looking at his reflection. The camera remains steady, capturing his movements and the mirrored background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: His overall movements aren't natural and realistic and his arms deform in the reflection and his reflection isn't consistent either. [Prompt Alignment]: He isn't holding a phone in his right hand, but in his reflection he did, though it still fails since he should be holding one and not his reflection.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/929.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: Two young men sit side by side at a kitchen counter with an open pizza box in front of them. Each man picks up a slice of pizza and takes a bite. The background shows kitchen cabinets and a bright countertop. The camera remains steady, capturing their actions and the casual kitchen setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The cause and effect of the bite is unnatural. [Prompt Alignment]: It is not shown in the video that they pick ups a slice of pizza\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/930.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Caucasian woman blows a bubble with her gum, popping it playfully before continuing to chew..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Her overall movements are unnatural. The effects of bubble gum are unrealistic. [Prompt Alignment]: She didn't pop the bubble gum.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/931.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Black man uses a mop to clean the tile floor, making sure the surface is free of dirt and spills..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Good [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/932.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The camera slowly and steadily zooms in on the Bosch power tools, focusing on their sturdy design and textured handles as they rest on a light surface..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The wordings are garbage symbols. [Motion & Physical Consistency]: Good. [Prompt Alignment]: Good.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/933.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A serene outdoor sports complex at midday, where an elderly East Asian woman clad in glasses and vibrant athletic gear laughs while cautiously passing a ball with her narrow, lightweight wooden field hockey stick..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The bouncing effect is not accurate and the ball looks floating. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/934.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged man with short dark hair, wearing a casual button-up shirt, stands in a cozy kitchen with warm lighting and wooden cabinets. He holds a small packet of seasoning powder in his right hand and sprinkles the powder directly onto a plate of spiced meat in front of him. The camera remains steady, showing the man’s actions and the inviting kitchen background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the cause-effect as the hand was keeps dispensing a seasoning powder on the plate and it was too much as the amount being dispense was huge making it appear unnatural and unrealistic. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/935.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The camera slowly glides across a dark room decorated with pumpkins and flickering candles. The background shows rows of pumpkins arranged on tables and scattered candles casting warm light. The camera movement is smooth and steady, capturing the glowing candles and the autumn-themed decor..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/936.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A clear iced coffee cup with a straw and visible condensation beads sits on a café tray. The cup is sliced cleanly in half, revealing several fluffy cake layers inside, all covered by a smooth, coffee-colored fondant. The background shows a wooden tabletop and blurred café elements. The camera remains steady, focusing on the cross-section of the cup and its cake interior..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The cake moves unnaturally without a cause. [Prompt Alignment]: The cup isn't sliced in half.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/937.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A plate of golden croissants and fresh fruit sits on a linen-covered table. The camera focuses closely on the flaky texture of the croissants, showing crisp layers and detail. The background includes the soft folds of the linen and hints of the surrounding tabletop. The shot remains steady, highlighting the pastries and fruit in natural light..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/938.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young East Asian man with short hair wears sunglasses that reflect the lights of a bustling nighttime market. Standing in the crowd, he tilts his head slightly, his steady pose exuding a composed vibe as pinks and blues from the surrounding lights illuminate the scene..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/939.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged South Asian man stands at a kitchen counter. He holds a bottle of cola in his right hand and steadily pours the dark liquid into a tall glass, watching as the bubbles rise. The background shows kitchen cabinets and a window with daylight. The camera remains steady, focusing on the man’s hands and the glass as he fills it..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: He holds a bottle of cola in his left hand.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/940.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Middle Eastern man with dark curly hair and a warm beige sweater lounges in a minimalistic home studio, pointing a black remote controller at a large projector screen. The screen flickers as it lowers smoothly into position, revealing an abstract presentation slide..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The object movement is not natural. There is also a cause-effect failure. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: The screen did not lower but instead it went up. The object movement is not natural. There is also a cause-effect failure..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/941.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Spanish woman with long dark hair sits on a velvet sofa in a softly lit living room. She wears a gold metallic luxury nightdress that shimmers gently under the light. Holding a wine glass in her right hand, she slowly turns her wrist, observing the sparkles on her outfit as she glances around the room. The camera captures her in crisp HD from the front, emphasizing the cinematic quality and the elegant setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Her left hand is deformed. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/942.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged Middle Eastern man in business attire enters a modern office, holding a folder in his right hand. He walks toward a large desk cluttered with files and a laptop. Behind the desk, a White male manager stands and points toward the door, causing the employee to look confused and glance between the manager and the exit. The employee pauses, lowers his eyes, then turns and walks slowly out of the frame as the manager watches. The camera remains steady, capturing the interaction and the office environment..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: man in background has texture issues [Motion & Physical Consistency]: hands are deformed [Prompt Alignment]: men look white rather than Middle Eastern. man is holding a folder in his left hand. no laptop on desk\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: No. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/943.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An older Black woman with silver-streaked hair and a red scarf sits on a large touring motorcycle parked near the edge of a cliff. She grips the handlebars and looks intently ahead, her gaze fixed on the sprawling desert landscape below, illuminated by golden morning light. The background shows rocky terrain and open desert. The camera remains steady, capturing her focused expression and the dramatic view..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: No [Motion & Physical Consistency]: UGC ( User Generated Content) Style - (Select UGC Style if it feels like a casual vlog or social clip) [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes - The video does not contain any nudity or harmful content. Cause-Effect: No - It looks like it was generated by Al. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/944.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: Two Black teens bump their smoothie bottles together in a playful cheers, grinning as they each take a long sip on a warm afternoon..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Movement of the fist appears stiff and not smooth. There is distortion on the bottle in the right frame of the screen. Stabilize to make it look normal. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/945.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: In a quiet office, a middle-aged British woman opens a packet of biscuits. The camera zooms in on her hand as she opens the packet, showing the crisp, sweet biscuits inside, perfect for a mid-afternoon break..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage [Motion & Physical Consistency]: The movement of the subject and the hand is not smooth and looks stiff. The fingers also appears to pass through the packet. The packet of biscuits suddenly open itself for no reason. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: The subject is not middle-aged. She did not open it. The movement of the subject and the hand is not smooth and looks stiff. The fingers also appears to pass through the packet. The packet of biscuits suddenly open itself for no reason..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/946.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: On a quiet path in Central Park, a young Black woman walks in a floral dress and skirt, surrounded by green trees and blooming flowers. The camera focuses on her movements as she walks leisurely, enjoying the calm and beauty of the park..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The movement of the woman while walking is stiff. There is also a subject-subject interaction issue as her face was deformed. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/947.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The video begins with a wide overhead view of a vibrant green lawn outdoors. The camera rapidly zooms in on a tall, clear glass set on a wooden table. A metal tap dispenses golden beer into the glass, and the foamy head rises with bubbles gathering inside. Foam spills slightly over the rim as the glass fills. The camera transitions to a medium shot, smoothly panning to keep both the tap and the glass clearly visible as the beer flows and the foam accumulates..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The texture of the grass is flickering. [Motion & Physical Consistency]: The cause-effect of the dripping beer is unnatural. [Prompt Alignment]: The object is not accurately mentioned as the beer foam did not spills slightly over the rim .\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/948.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black man with short, textured hair stands confidently in front of a full-length mirror inside a sleek, modern walk-in closet. He is wearing a snug blue athletic zip-up jacket with neon green stripes, paired with fitted black workout pants and white sneakers. Holding a smartphone in his right hand, he captures his reflection while his left hand adjusts the jacket collar to emphasize its unique design. The spacious closet, lined with organized shelves and soft lighting, adds a touch of sophistication to the scene..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: There is a flickering issue with the hanger. [Motion & Physical Consistency]: Unnatural hand movement. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: I think it's generated by AI because of the unnatural movement..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/949.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The subject, a South Asian man, uses a welding torch, his focused gaze behind the protective visor as the sparks fly..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: NULL [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/950.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A medium shot shows a sleek silver laptop resting on a wooden desk in a cozy home office. The laptop screen displays a financial calculator interface with monthly payment figures and input fields. A young Asian woman in a casual blue sweater sits at the desk, using her right hand to scroll and tap on the laptop’s touchpad, adjusting the numbers in the calculator. She pauses to read the updated monthly payment displayed on the screen. The camera slowly pans left, revealing more of the workspace and her focused posture..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage texts. [Motion & Physical Consistency]: There is a cause effect issue as the screen was moving even though the woman was not moving the touchpad. [Prompt Alignment]: The subject can't be determined due to lack of view. The content can't be determined if it was a financial calculator interface with monthly payment figures and input fields. She was not scrolling and tapping on the laptop’s touchpad, adjusting the numbers in the calculator.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: No. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/951.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young White woman with long black hair and an average build sits in the driver’s seat of a car, wearing a light-colored blouse. The car’s interior is clearly visible, with glimpses of the outside environment through the windows. She faces the camera, maintaining steady eye contact while speaking in a calm and friendly manner. Her facial expression is open and inviting. The camera remains fixed, capturing her natural delivery and the car’s interior..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: blurry eyebrows. [Motion & Physical Consistency]: good [Prompt Alignment]: subject looks asian and not white\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/952.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young adult college student with short dark hair, wearing a gray hoodie and jeans, stands beside a messy bed in a dorm room. The student holds a closed laptop in one hand and drops it onto the bed, where it lands among scattered notebooks and tangled sheets. The laptop’s lid bounces slightly upon impact. The background shows a cluttered desk and posters on the wall. The camera remains steady, capturing the action and the disorganized setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage symbols [Motion & Physical Consistency]: The cause-effect of the way he holds the laptop is unnatural. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/953.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A South Asian man in a casual button-up shirt stands in a softly lit living room. He holds a small candle in one hand and uses a lighter in his other hand to ignite the candle’s wick. The candle’s flame creates a warm glow that gently illuminates his face and the nearby furniture. The camera remains steady, capturing the man’s action and the cozy atmosphere..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/954.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged white man with a slim build and short hair sits against a neutral-colored wall, with a door frame visible on the left. He wears a white button-up shirt. He maintains steady eye contact with the camera, his facial expression subtly showing astonishment. The atmosphere is casual and intimate, focusing on his upper body and face. The camera remains fixed as he looks directly at the viewer..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: slight texture issues with his hair [Motion & Physical Consistency]: good [Prompt Alignment]: good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/955.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged East Asian woman sits in the passenger seat of a compact car illuminated by soft light. Her nails shimmer faintly as she gestures and talks sharply, emphasizing her intonation. Through the car’s window, a dark night sky is visible in the background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the subject movement as the hand of the subject was having stiffness and distortion. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: The subject is elderly. .\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/956.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A stylish Black woman poses confidently on a cobblestone street in SoHo, framed by cast-iron buildings and large windows, her trench coat flowing in the breeze..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Good. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/957.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A teenager with deep brown skin and braided hair crouches beside a sheep in a colorful meadow blooming with yellow and purple wildflowers. The sheep’s fluffy white wool falls away as she expertly snips with orange-handled shears under the soft morning light..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the subject interaction as the hand was deformed and passed through the scissors. Lastly, with the cause-effect, the subject was not firmly gripping the scissors, yet it was still able to cut the sheep's hair, which appeared unrealistic. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/958.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Caucasian woman with shoulder-length hair walks along a city sidewalk, holding a phone in her right hand. She speaks into the phone while moving forward, her left hand resting at her side. The background shows storefronts and passing pedestrians under daylight. The camera remains steady, capturing her conversation and natural walking motion..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the subject interaction as the body of the subject in the background was having a deformation upon moving. [Prompt Alignment]: The subject is not Caucasian.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/959.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A spatula glides through a layer of white whipped cream on a vibrant background, leaving perfect trails in its path..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: There is a flickering issue with the flower. [Motion & Physical Consistency]: The hand movement and the cause and effect of spreading are unnatural. [Prompt Alignment]: There is a hand in the video which is not mentioned in the prompt.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/960.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A group of men in their 30s huddle on a cozy living room couch, laughing and playfully shouting while battling each other in a competitive console game..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The movement of the subjects in the video while  laughing and playfully shouting appears stiff. There is a subject-subject interaction issue as the arms of the subjects are passing through each other. There is also a subject-object interaction issue as the console they are holding is passing through them. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/961.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The person gently holds up a nude-colored body shaper, subtly moving it forward to highlight its shape and design against a plain background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Good. [Prompt Alignment]: Good.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/962.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young woman with long dark hair sits at a vanity table in a softly lit room with pink theme filled with various makeup products. She holds an open lipstick in her right hand and uses it to swatch several different shades along her left forearm, revealing the colors in a neat row. The vanity surface is covered with makeup brushes and palettes, and a large mirror reflects the scene. The camera remains steady, capturing her focused actions and the whole room..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage texts. [Motion & Physical Consistency]: There is a subject interaction issue as the hand of the woman was deformed, There is also a cause effect issue as the lipstick did not leave visible marks. [Prompt Alignment]: She is holding the lipstick on her left hand.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/963.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A group of Latino and Black restaurant workers operate together in a lively, brightly lit dining area at night. In a medium shot, two men in white dress shirts and aprons walk past decorated tables with colorful lights, carrying trays of drinks. Nearby, a woman in a red dress shirt arranges plates and smiles at seated guests. The camera slowly pans across the scene, showing staff moving between tables to clear dishes and serve food. The background features festive decorations and warm lighting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The plate the woman in red is holding fuses together with the other plate as well as the hand of the man serving a drink, his left hand deforms as well and the effects of an object that floats and unnaturally disappears. [Prompt Alignment]: There were no trays of drinks and they weren't carrying it either.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/964.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman plays a soft melody on her flute, her fingers delicately moving over the keys as the notes float into the air..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: NULL [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/965.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: In a vibrant city square, a Middle Eastern man with a ponytail runs in black sports shoes. The camera zooms in on his lower body as he quickly navigates through the crowd, showing off the shoes' comfort and agility on the city pavement..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the subject movement as the subject was running with stiffness and not smooth. Also with the subject interaction as the feet was having a deformation. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/966.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: Under a vibrant aurora in a snowy Arctic landscape, a young South Asian man in a thick parka and his elderly Latina friend with silver hair hold steaming glasses of hot cocoa. Their breath fogs lightly in the icy air as they beam at each other and clink their mugs before sipping, the swirling northern lights creating a magical, prismatic glow across the snowy terrain..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Hand appearance is not normal and there is bubbles in the cup. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: The subject is not South Asian and the breath is not come from their mouth..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/967.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged Latino man, dressed in a rugged, weathered leather jacket and dark jeans, stands in a rocky desert landscape. Striking a side pose with his gaze focused on the horizon, his hand rests by his side while his other hand pushes back his dark, slightly windswept hair. The warm hues of the setting sun emphasize the rough textures of his jacket and the rugged terrain around him..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Good. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/968.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A close-up shot captures a hand placing an empty ceramic mug under a coffee machine spout; dark coffee begins to stream in, rising and forming crema at the top..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The hand movement is unnatural. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: There is a pitcher of some sort pouring coffee into the mug. The ceramic mug was not put under the coffee machine,.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/969.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman with shoulder-length, styled hair stands in a well-lit bathroom. She wears a fitted T-shirt. Holding a can of hair spray at arm’s length in her right hand, she evenly sprays her hair, turning her head slightly to cover all sides. The background shows a bathroom mirror and countertop. The camera remains steady, capturing her focused motion and the setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage symbols, not English. [Motion & Physical Consistency]: The deformed hand and the cause and effect of the sudden appearance of the other bottle seems unnatural. [Prompt Alignment]: It is not a bathroom mirror and she is holding two cans  of hair spray at her both hands.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/970.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Black woman uses her fingers to gently dab and blend cream blush onto the apples of her cheeks, creating a soft, rosy glow..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The movement of the woman while dabbing the blush is not natural as it is too fast. There is also a cause effect issue as the blush was not blended on every dab she makes. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: The movement of the woman while dabbing the blush is not natural as it is too fast. There is also a cause effect issue as the blush was not blended on every dab she makes. I believe the video is AI-generated because of this discrepancy..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/971.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The video begins with a close-up shot of two Vaseline Pro Derma body lotion bottles. The first bottle is labeled \"4\" and is described as a moisturizing lotion with hyaluronic acid, intended for dry skin. The second bottle is labeled \"3\" and is a brightening lotion with niacinamide, aimed at dull and uneven skin. Both bottles are white with metallic pump dispensers, and the labels are prominently displayed in blue and pink colors respectively.  The camera angle is eye-level, capturing the bottles from a side view. The shot focuses on the text and branding on the bottles, with the words \"MY FAVORITES\" and \"Vaseline Pro Derma Body Lotions\" overlaid on the image.  In the following frame, a hand is shown in close-up, with a lotion pump being pressed to dispense a small amount of lotion onto the skin. The hand appears to belong to a male, with visible hair and a slightly tanned complexion. The lotion is white and creamy, forming a small dollop on the back of the hand.  The camera angle remains at eye-level, capturing the hand from a side view as the lotion is applied. Text overlays describe the lotion's features, including Pro Ceramide Technology, its ability to repair, moisturize, and strengthen the skin barrier, and its suitability for sensitive skin. The background is neutral, with soft lighting highlighting the lotion application process..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Garbage symbols, not English. [Motion & Physical Consistency]: Good [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: The object in the video has garbage symbols, so we cannot identify whether it is Vaseline Pro Derma..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/972.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The subject, a South Asian woman, eats nachos with sour cream, each bite refreshing as she enjoys the tangy flavors..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: NULL [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/973.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young Asian woman with glowing, smooth skin sits in a softly lit bedroom. She wears a white tank top and has a relaxed, refreshed expression. Sunlight streams through sheer curtains, casting gentle shadows on the room’s furnishings. The camera remains steady, focusing on her face and upper body as she gently shifts her posture, maintaining her calm, healthy appearance..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Good. [Prompt Alignment]: Good.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/974.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Latina teenager with long dark hair stands by a window in a sunlit bedroom. She wears a casual T-shirt. She dispenses a small amount of lightweight moisturizer onto her fingertips, then uses both hands to gently apply and blend the moisturizer onto her cheeks and nose in smooth, even motions. The background shows soft daylight streaming in and simple bedroom decor. The camera remains steady, focusing on her face and hands as she completes the skincare routine..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: The moisturizer was not spread accurately. [Prompt Alignment]: She didn't dispense as the moisturizer was already in her hands.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/975.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly South Asian woman with silver-streaked hair neatly tied into a bun admires herself in a large, ornately framed mirror in her walk-in closet. She’s wearing a cream-colored cropped jacket adorned with fine silver embroidery, paired with loose linen pants. Her left hand gracefully touches the gold bracelet on her wrist while she lightly adjusts the jacket hem. The soft glow from the warm closet lighting highlights her elegant appearance as she captures her reflection on her phone..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: There is garbage text visible in the video. [Motion & Physical Consistency]: The woman's movement is unnatural, with her hand moving excessively. There is also a subject-object interaction issue with how she holds the phone, and her other foot appears to pass through the mirror. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Her left hand did not gracefully touch the gold bracelet on her wrist. The woman's movement is unnatural, with her hand moving excessively. There is also a subject-object interaction issue with how she holds the phone, and her other foot appears to pass through the mirror. I believe the video is AI-generated because of this discrepancy..\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/976.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A South Asian man with medium brown skin stands in a well-lit bathroom, facing the mirror. He holds a metal applicator in his right hand and gently glides it under his slightly puffy lower eyelids, applying a cooling under-eye gel. His left hand rests by his side. The background shows a bathroom counter with toiletries and a tiled wall. The camera remains steady, capturing his focused application and the calm setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: His overall movements are unnatural. Interaction with the metal applicator is unrealistic and unusual. The effects of the cream is also unrealistic when applied to his under-eye. [Prompt Alignment]: It doesn't look the same as the real metal applicator and he isn't holding it to his right hand.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/977.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A white man with short hair and glasses sits in an office chair at his desk, illuminated by the soft glow of his laptop screen. He leans back in the chair, casually scrolling on the laptop with one hand while quietly chuckling. The background shows a tidy home office with shelves and warm ambient lighting. The camera remains steady, focusing on his relaxed posture and subtle laughter..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is a cause effect issue as the man was typing but there is no changes on the monitor. [Prompt Alignment]: He does not appear scrolling through his laptop.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/978.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young white woman with long blonde hair in loose waves stands confidently against a minimalist urban backdrop with clean lines and muted tones. She wears an oversized graphic sweatshirt, high-waisted cargo pants, and chunky sneakers. Natural light softly illuminates her face as she looks directly at the camera and speaks, maintaining steady eye contact. Her expression is calm and subtly relieved, as if she is sharing positive insights about fashion. The camera remains steady, capturing her streetwear style and relaxed, authentic demeanor..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: The chunky shoes aren't visible and she maintains steady eye contact.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/979.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A filipino with long, wavy hair is seen applying makeup with a brush in her right hand, wearing a black off-shoulder top and a gold bracelet on her left wrist. She continues to apply makeup, using a smaller brush to blend the product on her face. The woman then uses a larger brush to apply makeup to her cheeks, focusing on even coverage..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is a cause effect issue as the brush did not have marks and the product was not blended. [Prompt Alignment]: The bracelet is on her right wrist. There is no larger brush visible.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/980.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly East Asian man with short gray hair stands in a quiet, book-lined room. He presses a button on the side of his desk, raising it to a standing height. Once the desk is elevated, he begins sorting papers on its surface with both hands. The background shows shelves filled with books and soft, natural lighting. The camera remains steady, capturing his actions and the peaceful setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is a subject interaction issue as his hand was deformed. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/981.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young East Asian girl and a young East Asian man stand on opposite sides of a table tennis table in a brightly lit indoor sports facility. Both hold table tennis paddles and rally the ball back and forth across the net, each taking quick turns to return the ball. The background shows other tables and sports equipment. The camera remains steady, capturing their coordinated movements and the active setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: The movement is not smooth and looks stiff. The hand and object seems deformed and the ball seems to pass through the net. The ball seems to disappear, float, and bounce unnaturally. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/982.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young South Asian woman in a lush green park sits cross-legged with a digital tablet on her lap, using a stylus to paint a digital landscape of a mountain, her face glowing from the device's screen in the evening light..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: She is sketching but it didn't sketch on the tablet, which is unrealistic. [Prompt Alignment]: She isn't South Asian.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/983.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A man holds a black leather shoe, slowly rotating it to showcase its design and sole, while standing in front of a wooden wardrobe with hanging clothes and shelves..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: His left hand fingers are deformed. [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/984.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Hispanic woman with long dark hair stands at a vanity table in a softly lit room. She holds a bronzer compact in one hand and uses her other hand to lift open the lid, revealing the finely pressed powder inside. She tilts the compact slightly so the light highlights the powder’s surface. The camera remains steady, focusing on her hands and the compact against the vanity background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The deformed bronzer compact seem unnatural. [Prompt Alignment]: The vanity is not visible. She holds a bronzer compact with both hands.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/985.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A South Asian woman in a casual T-shirt stands in a bright kitchen. She holds a slice of pizza in one hand and a sheet of parchment paper in the other. She wraps the pizza slice in the parchment paper, folding the edges neatly around it. The background shows kitchen counters and a refrigerator. The camera remains steady, focusing on her hands as she wraps the pizza..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Her hand got deformed. [Prompt Alignment]: She isn't South Asian.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/986.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Black man cleans his shoes with a soft brush, the polish restoring their shine as he works carefully..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: NULL [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/987.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Black man uses a traditional razor to shave his face, paying close attention to his facial contours for a clean shave..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The movement is not smooth and looks stiff. The object seems deformed. The cream did not spread accurately. [Prompt Alignment]: It is not a traditional razor.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: No. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/988.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A woman with curly blond hair works at a simple desk, using a mouse and typing on a keyboard. Her movements are steady and intentional, focusing on a task displayed on a screen in front of her. The environment is quiet, with faint background sounds of machinery creating a calm atmosphere..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: The subject's hand is distorted and the movement is unnatural. She tried to type the keyboard but her hand did not touch it.  Stabilize to make it look natural. [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: No. Physical Interaction: No. Cause-Effect: No. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/989.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: The subject, a South Asian man, plays a classical piece on the flute, his breath steady as the music flows from the instrument..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: NULL [Motion & Physical Consistency]: Good [Prompt Alignment]: NULL\n</think>\n<answer>\nVideo Quality: NULL. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: NULL. Object Existence: NULL. Subject-Object Interaction: NULL.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/990.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A white man with light stubble stands in a softly lit bedroom. He holds a folded hoodie in both hands, lifts it to his nose, and closes his eyes as he inhales the fresh laundry scent. The camera remains steady, capturing his relaxed expression and the cozy bedroom background..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: He is already closing his eyes from the start of the video.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/991.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A wide shot shows a two-story European-style house with a sloped red roof and light-colored walls on a quiet residential street. The camera begins by capturing the clean roof without visible solar panels, then smoothly pans to reveal several dark-colored solar panels installed on the roof. Three workers in blue coveralls stand on the rooftop beside the panels, facing the camera and giving a thumbs-up gesture. The background includes trees and neighboring houses. The camera continues its steady pan across the scene..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: There is an issue with the subject interaction as the faces of the subject was seen having a deformation. [Prompt Alignment]: Four worker standing on the roof.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/992.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young White woman with blue eyes looks into a mirror as she applies mascara to her upper eyelashes using a mascara wand. The camera captures a crisp close-up of her face and the precise motion of the wand as she coats her lashes. The background is softly blurred, emphasizing her eyes and the makeup application..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Texture on the mascara wand. [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/993.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A middle-aged South Asian woman wearing a vibrant turquoise workout outfit sits on a stability ball in a modern gym. She repeatedly performs slow situps, bending her knees and leaning her upper body forward, then returning to a seated position on the ball. Her movements are deliberate and controlled. The background shows gym equipment and bright lighting. The camera remains steady, capturing her exercise form and the gym environment..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Deformed hand is unnatural. [Prompt Alignment]: She's not a South Asian woman.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: Yes. Subject Existence: No. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/994.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A close-up shot shows four teenage girls seated together on a cozy couch, each wearing matching pajamas. The camera alternates between close-ups of each girl as she interacts with her phone. In Cut 1, Girl 1 holds a phone in her hands. In Cut 2, Girl 2 uses her thumb to tap the calculator app on her phone. In Cut 3, Girl 3 scrolls through dessert options on her phone. In Cut 4, Girl 4 displays the final order on her screen. The camera then cuts to all four girls tapping the \"Order Now\" button on their phones at the same time. The background features a softly lit living room with pillows and a coffee table..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: The wordings are garbage symbols. [Motion & Physical Consistency]: Interaction with the object is not in accordance to physical law. Cause-effect failure as the changes on the phone display are not accurate. [Prompt Alignment]: There are no four teenagers. Can't determine if it's a calculator app. The camera didn't alternate between close-ups of each girl as she interacts with her phone.\n</think>\n<answer>\nVideo Quality: No. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: No. Object Existence: No. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/995.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A Middle Eastern woman with long dark hair stands in a well-lit bedroom. She holds a folded sweater close to her face with both hands and gently sniffs the fabric. Her eyes close briefly as she inhales, and she smiles softly. The background shows a neatly made bed and a window with sunlight streaming in. The camera remains steady, capturing her relaxed posture and subtle expression..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Good [Prompt Alignment]: Good\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: Yes.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/996.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A young woman with straight black hair stands in a bright living room. She holds a cardboard shipping box in both hands and lifts the lid to open it, revealing several items inside. The background shows a sofa and a coffee table. The camera remains steady, capturing her actions and the cozy room setting..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: Deformed hand and the cause-effect of the disappearance of the lid of the box is unnatural. [Prompt Alignment]: The box is already open.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: No. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/997.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: A playful Golden Retriever eagerly eats its dog food from a bowl in a cozy living room. The camera captures the dog's joyful face as it munches on the food, with colorful toys scattered around it..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good. [Motion & Physical Consistency]: Cause-effect failure as some of the dog food suddenly disappears. [Prompt Alignment]: Yes\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: Yes. Physical Interaction: Yes. Cause-Effect: No. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: I think it's generated by AI\n.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/998.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: An elderly East Asian man with a calm demeanor stands in an intricately designed Zen garden, surrounded by cherry blossoms gently falling in slow motion. He holds a shimmering, metallic orb emitting soft, pulsating light as he tilts it gently, watching ripples of energy create mesmerizing waves through the air..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Good [Motion & Physical Consistency]: The movement of the man appears stiff. [Prompt Alignment]: There are no cherry blossoms gently falling in slow motion. The man is not tilting the orb.\n</think>\n<answer>\nVideo Quality: Yes. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: No.\n</answer>\n"
            }
        ]
    },
    {
        "videos": [
            "videos/eval_0/999.mp4"
        ],
        "conversations": [
            {
                "from": "human",
                "value": "<video>\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos.\nPlease watch the frames of a given video.\nEvaluate the video according to the following three dimensions.\n\n[Visual Quality]\nAssess the video in terms of:\nVideo Quality: whether the video is free from major visual defects, including blur, lack of detail, poor texture, lighting issues, color distortion, flickering, and overexposure.\n\n[Motion & Physical Consistency]\nAssess the video in terms of:\nSubject Movement: whether the subject's motion is natural, smooth, and physically realistic.\nPhysical Interaction: whether interactions among subjects and/or objects are physically plausible.\nCause-Effect: whether causal relationships are correctly depicted.\n\n[Prompt Alignment]\nTextual prompt: Inside a dimly lit, industrial-style studio with spotlights above, a tall young man with dark brown skin and styled dreadlocks takes a deliberate step left, showcasing a sleek, fitted leather jacket. His arms rest naturally by his sides, the polished material reflecting soft highlights. As he turns slightly, his smooth, assured movements highlight the jacket's tailored fit and versatility..\nAssess whether the video is well-aligned with the textual prompt in terms of:\nSubject Existence: whether the subject described in the prompt appears and is accurate.\nObject Existence: whether the object described in the prompt appears and is accurate.\nSubject-Object Interaction: whether the interaction described in the prompt is correctly represented.\n\nProvide your reasoning, then output \"Yes\" or \"No\".\n"
            },
            {
                "from": "gpt",
                "value": "\n<think>\n[Visual Quality]: Flickering issue in the background. [Motion & Physical Consistency]: The subject's body movement is not normal. [Prompt Alignment]: No\n</think>\n<answer>\nVideo Quality: No. Subject Movement: No. Physical Interaction: Yes. Cause-Effect: Yes. Subject Existence: Yes. Object Existence: Yes. Subject-Object Interaction: The subject's first step is in the right..\n</answer>\n"
            }
        ]
    }
]