Datasets:

hexuan21
/

VideoScore_legacy

License:

hexuan21 commited on Sep 7, 2024

Commit

668a317

verified ·

1 Parent(s): 44fb77b

upload refined_40k/test_bench 's json, videos and frames

Browse files

Files changed (1) hide show

data/refined_40k/test_bench/data_bad_null_prompt.json +646 -0

data/refined_40k/test_bench/data_bad_null_prompt.json ADDED Viewed

	@@ -0,0 +1,646 @@

+[
+    {
+        "id": "e702640",
+        "images": [
+            "e702640_00.jpg",
+            "e702640_01.jpg",
+            "e702640_02.jpg",
+            "e702640_03.jpg",
+            "e702640_04.jpg",
+            "e702640_05.jpg",
+            "e702640_06.jpg",
+            "e702640_07.jpg"
+        ],
+        "conversations": [
+            {
+                "from": "human",
+                "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"\",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> \n"
+            },
+            {
+                "from": "gpt",
+                "value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 2\n text-to-video alignment: 1\n factual consistency: 3\n"
+            }
+        ]
+    },
+    {
+        "id": "e402795",
+        "images": [
+            "e402795_00.jpg",
+            "e402795_01.jpg",
+            "e402795_02.jpg",
+            "e402795_03.jpg",
+            "e402795_04.jpg",
+            "e402795_05.jpg",
+            "e402795_06.jpg",
+            "e402795_07.jpg",
+            "e402795_08.jpg",
+            "e402795_09.jpg",
+            "e402795_10.jpg",
+            "e402795_11.jpg",
+            "e402795_12.jpg",
+            "e402795_13.jpg",
+            "e402795_14.jpg",
+            "e402795_15.jpg"
+        ],
+        "conversations": [
+            {
+                "from": "human",
+                "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"\",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
+            },
+            {
+                "from": "gpt",
+                "value": "visual quality: 2\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 2\n"
+            }
+        ]
+    },
+    {
+        "id": "e204933",
+        "images": [
+            "e204933_00.jpg",
+            "e204933_01.jpg",
+            "e204933_02.jpg",
+            "e204933_03.jpg",
+            "e204933_04.jpg",
+            "e204933_05.jpg",
+            "e204933_06.jpg",
+            "e204933_07.jpg",
+            "e204933_08.jpg",
+            "e204933_09.jpg",
+            "e204933_10.jpg",
+            "e204933_11.jpg",
+            "e204933_12.jpg",
+            "e204933_13.jpg",
+            "e204933_14.jpg",
+            "e204933_15.jpg"
+        ],
+        "conversations": [
+            {
+                "from": "human",
+                "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"\",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
+            },
+            {
+                "from": "gpt",
+                "value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 1\n"
+            }
+        ]
+    },
+    {
+        "id": "e403697",
+        "images": [
+            "e403697_00.jpg",
+            "e403697_01.jpg",
+            "e403697_02.jpg",
+            "e403697_03.jpg",
+            "e403697_04.jpg",
+            "e403697_05.jpg",
+            "e403697_06.jpg",
+            "e403697_07.jpg",
+            "e403697_08.jpg",
+            "e403697_09.jpg",
+            "e403697_10.jpg",
+            "e403697_11.jpg",
+            "e403697_12.jpg",
+            "e403697_13.jpg",
+            "e403697_14.jpg",
+            "e403697_15.jpg"
+        ],
+        "conversations": [
+            {
+                "from": "human",
+                "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"\",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
+            },
+            {
+                "from": "gpt",
+                "value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 2\n"
+            }
+        ]
+    },
+    {
+        "id": "e206780",
+        "images": [
+            "e206780_00.jpg",
+            "e206780_01.jpg",
+            "e206780_02.jpg",
+            "e206780_03.jpg",
+            "e206780_04.jpg",
+            "e206780_05.jpg",
+            "e206780_06.jpg",
+            "e206780_07.jpg",
+            "e206780_08.jpg",
+            "e206780_09.jpg",
+            "e206780_10.jpg",
+            "e206780_11.jpg",
+            "e206780_12.jpg",
+            "e206780_13.jpg",
+            "e206780_14.jpg",
+            "e206780_15.jpg"
+        ],
+        "conversations": [
+            {
+                "from": "human",
+                "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"\",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
+            },
+            {
+                "from": "gpt",
+                "value": "visual quality: 2\n temporal consistency: 3\n dynamic degree: 2\n text-to-video alignment: 1\n factual consistency: 1\n"
+            }
+        ]
+    },
+    {
+        "id": "e305808",
+        "images": [
+            "e305808_00.jpg",
+            "e305808_01.jpg",
+            "e305808_02.jpg",
+            "e305808_03.jpg",
+            "e305808_04.jpg",
+            "e305808_05.jpg",
+            "e305808_06.jpg",
+            "e305808_07.jpg",
+            "e305808_08.jpg",
+            "e305808_09.jpg",
+            "e305808_10.jpg",
+            "e305808_11.jpg",
+            "e305808_12.jpg",
+            "e305808_13.jpg",
+            "e305808_14.jpg",
+            "e305808_15.jpg"
+        ],
+        "conversations": [
+            {
+                "from": "human",
+                "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"\",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
+            },
+            {
+                "from": "gpt",
+                "value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 3\n"
+            }
+        ]
+    },
+    {
+        "id": "e501654",
+        "images": [
+            "e501654_00.jpg",
+            "e501654_01.jpg",
+            "e501654_02.jpg",
+            "e501654_03.jpg",
+            "e501654_04.jpg",
+            "e501654_05.jpg",
+            "e501654_06.jpg",
+            "e501654_07.jpg",
+            "e501654_08.jpg",
+            "e501654_09.jpg",
+            "e501654_10.jpg",
+            "e501654_11.jpg",
+            "e501654_12.jpg",
+            "e501654_13.jpg",
+            "e501654_14.jpg",
+            "e501654_15.jpg"
+        ],
+        "conversations": [
+            {
+                "from": "human",
+                "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"\",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
+            },
+            {
+                "from": "gpt",
+                "value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 1\n text-to-video alignment: 1\n factual consistency: 2\n"
+            }
+        ]
+    },
+    {
+        "id": "e004131",
+        "images": [
+            "e004131_00.jpg",
+            "e004131_01.jpg",
+            "e004131_02.jpg",
+            "e004131_03.jpg",
+            "e004131_04.jpg",
+            "e004131_05.jpg",
+            "e004131_06.jpg",
+            "e004131_07.jpg",
+            "e004131_08.jpg",
+            "e004131_09.jpg",
+            "e004131_10.jpg",
+            "e004131_11.jpg",
+            "e004131_12.jpg",
+            "e004131_13.jpg",
+            "e004131_14.jpg",
+            "e004131_15.jpg",
+            "e004131_16.jpg",
+            "e004131_17.jpg",
+            "e004131_18.jpg",
+            "e004131_19.jpg",
+            "e004131_20.jpg",
+            "e004131_21.jpg",
+            "e004131_22.jpg",
+            "e004131_23.jpg"
+        ],
+        "conversations": [
+            {
+                "from": "human",
+                "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"\",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
+            },
+            {
+                "from": "gpt",
+                "value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 2\n text-to-video alignment: 1\n factual consistency: 2\n"
+            }
+        ]
+    },
+    {
+        "id": "e602058",
+        "images": [
+            "e602058_00.jpg",
+            "e602058_01.jpg",
+            "e602058_02.jpg",
+            "e602058_03.jpg",
+            "e602058_04.jpg",
+            "e602058_05.jpg",
+            "e602058_06.jpg",
+            "e602058_07.jpg",
+            "e602058_08.jpg",
+            "e602058_09.jpg",
+            "e602058_10.jpg",
+            "e602058_11.jpg",
+            "e602058_12.jpg",
+            "e602058_13.jpg",
+            "e602058_14.jpg",
+            "e602058_15.jpg"
+        ],
+        "conversations": [
+            {
+                "from": "human",
+                "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"\",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
+            },
+            {
+                "from": "gpt",
+                "value": "visual quality: 3\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 3\n"
+            }
+        ]
+    },
+    {
+        "id": "e003968",
+        "images": [
+            "e003968_00.jpg",
+            "e003968_01.jpg",
+            "e003968_02.jpg",
+            "e003968_03.jpg",
+            "e003968_04.jpg",
+            "e003968_05.jpg",
+            "e003968_06.jpg",
+            "e003968_07.jpg",
+            "e003968_08.jpg",
+            "e003968_09.jpg",
+            "e003968_10.jpg",
+            "e003968_11.jpg",
+            "e003968_12.jpg",
+            "e003968_13.jpg",
+            "e003968_14.jpg",
+            "e003968_15.jpg",
+            "e003968_16.jpg",
+            "e003968_17.jpg",
+            "e003968_18.jpg",
+            "e003968_19.jpg",
+            "e003968_20.jpg",
+            "e003968_21.jpg",
+            "e003968_22.jpg",
+            "e003968_23.jpg"
+        ],
+        "conversations": [
+            {
+                "from": "human",
+                "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"\",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
+            },
+            {
+                "from": "gpt",
+                "value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 2\n text-to-video alignment: 1\n factual consistency: 1\n"
+            }
+        ]
+    },
+    {
+        "id": "e305403",
+        "images": [
+            "e305403_00.jpg",
+            "e305403_01.jpg",
+            "e305403_02.jpg",
+            "e305403_03.jpg",
+            "e305403_04.jpg",
+            "e305403_05.jpg",
+            "e305403_06.jpg",
+            "e305403_07.jpg",
+            "e305403_08.jpg",
+            "e305403_09.jpg",
+            "e305403_10.jpg",
+            "e305403_11.jpg",
+            "e305403_12.jpg",
+            "e305403_13.jpg",
+            "e305403_14.jpg",
+            "e305403_15.jpg"
+        ],
+        "conversations": [
+            {
+                "from": "human",
+                "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"\",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
+            },
+            {
+                "from": "gpt",
+                "value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 1\n text-to-video alignment: 1\n factual consistency: 2\n"
+            }
+        ]
+    },
+    {
+        "id": "e603064",
+        "images": [
+            "e603064_00.jpg",
+            "e603064_01.jpg",
+            "e603064_02.jpg",
+            "e603064_03.jpg",
+            "e603064_04.jpg",
+            "e603064_05.jpg",
+            "e603064_06.jpg",
+            "e603064_07.jpg",
+            "e603064_08.jpg",
+            "e603064_09.jpg",
+            "e603064_10.jpg",
+            "e603064_11.jpg",
+            "e603064_12.jpg",
+            "e603064_13.jpg",
+            "e603064_14.jpg",
+            "e603064_15.jpg"
+        ],
+        "conversations": [
+            {
+                "from": "human",
+                "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"\",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
+            },
+            {
+                "from": "gpt",
+                "value": "visual quality: 1\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 1\n"
+            }
+        ]
+    },
+    {
+        "id": "e206274",
+        "images": [
+            "e206274_00.jpg",
+            "e206274_01.jpg",
+            "e206274_02.jpg",
+            "e206274_03.jpg",
+            "e206274_04.jpg",
+            "e206274_05.jpg",
+            "e206274_06.jpg",
+            "e206274_07.jpg",
+            "e206274_08.jpg",
+            "e206274_09.jpg",
+            "e206274_10.jpg",
+            "e206274_11.jpg",
+            "e206274_12.jpg",
+            "e206274_13.jpg",
+            "e206274_14.jpg",
+            "e206274_15.jpg"
+        ],
+        "conversations": [
+            {
+                "from": "human",
+                "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"\",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
+            },
+            {
+                "from": "gpt",
+                "value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 3\n"
+            }
+        ]
+    },
+    {
+        "id": "e803247",
+        "images": [
+            "e803247_00.jpg",
+            "e803247_01.jpg",
+            "e803247_02.jpg",
+            "e803247_03.jpg",
+            "e803247_04.jpg",
+            "e803247_05.jpg",
+            "e803247_06.jpg",
+            "e803247_07.jpg",
+            "e803247_08.jpg",
+            "e803247_09.jpg",
+            "e803247_10.jpg",
+            "e803247_11.jpg",
+            "e803247_12.jpg",
+            "e803247_13.jpg",
+            "e803247_14.jpg",
+            "e803247_15.jpg",
+            "e803247_16.jpg",
+            "e803247_17.jpg",
+            "e803247_18.jpg",
+            "e803247_19.jpg",
+            "e803247_20.jpg",
+            "e803247_21.jpg",
+            "e803247_22.jpg",
+            "e803247_23.jpg"
+        ],
+        "conversations": [
+            {
+                "from": "human",
+                "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"\",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
+            },
+            {
+                "from": "gpt",
+                "value": "visual quality: 2\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 1\n"
+            }
+        ]
+    },
+    {
+        "id": "e105457",
+        "images": [
+            "e105457_00.jpg",
+            "e105457_01.jpg",
+            "e105457_02.jpg",
+            "e105457_03.jpg",
+            "e105457_04.jpg",
+            "e105457_05.jpg",
+            "e105457_06.jpg",
+            "e105457_07.jpg",
+            "e105457_08.jpg",
+            "e105457_09.jpg",
+            "e105457_10.jpg",
+            "e105457_11.jpg",
+            "e105457_12.jpg",
+            "e105457_13.jpg",
+            "e105457_14.jpg",
+            "e105457_15.jpg"
+        ],
+        "conversations": [
+            {
+                "from": "human",
+                "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"\",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
+            },
+            {
+                "from": "gpt",
+                "value": "visual quality: 2\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 3\n"
+            }
+        ]
+    },
+    {
+        "id": "e802445",
+        "images": [
+            "e802445_00.jpg",
+            "e802445_01.jpg",
+            "e802445_02.jpg",
+            "e802445_03.jpg",
+            "e802445_04.jpg",
+            "e802445_05.jpg",
+            "e802445_06.jpg",
+            "e802445_07.jpg",
+            "e802445_08.jpg",
+            "e802445_09.jpg",
+            "e802445_10.jpg",
+            "e802445_11.jpg",
+            "e802445_12.jpg",
+            "e802445_13.jpg",
+            "e802445_14.jpg",
+            "e802445_15.jpg",
+            "e802445_16.jpg",
+            "e802445_17.jpg",
+            "e802445_18.jpg",
+            "e802445_19.jpg",
+            "e802445_20.jpg",
+            "e802445_21.jpg",
+            "e802445_22.jpg",
+            "e802445_23.jpg"
+        ],
+        "conversations": [
+            {
+                "from": "human",
+                "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"\",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
+            },
+            {
+                "from": "gpt",
+                "value": "visual quality: 1\n temporal consistency: 1\n dynamic degree: 2\n text-to-video alignment: 1\n factual consistency: 1\n"
+            }
+        ]
+    },
+    {
+        "id": "e106975",
+        "images": [
+            "e106975_00.jpg",
+            "e106975_01.jpg",
+            "e106975_02.jpg",
+            "e106975_03.jpg",
+            "e106975_04.jpg",
+            "e106975_05.jpg",
+            "e106975_06.jpg",
+            "e106975_07.jpg",
+            "e106975_08.jpg",
+            "e106975_09.jpg",
+            "e106975_10.jpg",
+            "e106975_11.jpg",
+            "e106975_12.jpg",
+            "e106975_13.jpg",
+            "e106975_14.jpg",
+            "e106975_15.jpg"
+        ],
+        "conversations": [
+            {
+                "from": "human",
+                "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"\",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
+            },
+            {
+                "from": "gpt",
+                "value": "visual quality: 2\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 3\n"
+            }
+        ]
+    },
+    {
+        "id": "e603626",
+        "images": [
+            "e603626_00.jpg",
+            "e603626_01.jpg",
+            "e603626_02.jpg",
+            "e603626_03.jpg",
+            "e603626_04.jpg",
+            "e603626_05.jpg",
+            "e603626_06.jpg",
+            "e603626_07.jpg",
+            "e603626_08.jpg",
+            "e603626_09.jpg",
+            "e603626_10.jpg",
+            "e603626_11.jpg",
+            "e603626_12.jpg",
+            "e603626_13.jpg",
+            "e603626_14.jpg",
+            "e603626_15.jpg"
+        ],
+        "conversations": [
+            {
+                "from": "human",
+                "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"\",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
+            },
+            {
+                "from": "gpt",
+                "value": "visual quality: 3\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 2\n"
+            }
+        ]
+    },
+    {
+        "id": "e305932",
+        "images": [
+            "e305932_00.jpg",
+            "e305932_01.jpg",
+            "e305932_02.jpg",
+            "e305932_03.jpg",
+            "e305932_04.jpg",
+            "e305932_05.jpg",
+            "e305932_06.jpg",
+            "e305932_07.jpg",
+            "e305932_08.jpg",
+            "e305932_09.jpg",
+            "e305932_10.jpg",
+            "e305932_11.jpg",
+            "e305932_12.jpg",
+            "e305932_13.jpg",
+            "e305932_14.jpg",
+            "e305932_15.jpg"
+        ],
+        "conversations": [
+            {
+                "from": "human",
+                "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"\",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
+            },
+            {
+                "from": "gpt",
+                "value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 1\n"
+            }
+        ]
+    },
+    {
+        "id": "e103509",
+        "images": [
+            "e103509_00.jpg",
+            "e103509_01.jpg",
+            "e103509_02.jpg",
+            "e103509_03.jpg",
+            "e103509_04.jpg",
+            "e103509_05.jpg",
+            "e103509_06.jpg",
+            "e103509_07.jpg",
+            "e103509_08.jpg",
+            "e103509_09.jpg",
+            "e103509_10.jpg",
+            "e103509_11.jpg",
+            "e103509_12.jpg",
+            "e103509_13.jpg",
+            "e103509_14.jpg",
+            "e103509_15.jpg"
+        ],
+        "conversations": [
+            {
+                "from": "human",
+                "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"\",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
+            },
+            {
+                "from": "gpt",
+                "value": "visual quality: 1\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 3\n"
+            }
+        ]
+    }
+]