Download inputs/shot_08/input.json from ymyy307/testltx: direct link, hf CLI and curl.
- Browser
- Download file 3.69 kB
-
https://huggingface.co/ymyy307/testltx/resolve/main/inputs/shot_08/input.json
- Command line
-
hf download hf://ymyy307/testltx/inputs/shot_08/input.json
-
curl -L -o input.json https://huggingface.co/ymyy307/testltx/resolve/main/inputs/shot_08/input.json
3.69 kB
| { | |
| "shot_index": 8, | |
| "raw_prompt": "<Subject 2> is an older East Asian male birdwatcher with neatly cut gray hair, a brown waterproof coat, and compact binoculars. At normal speed, <Subject 2> raises his binoculars toward the open sky rather than the reeds, follows one distant movement, and lowers them without pointing it out. No intelligible dialogue is heard. The shot uses realistic cinematic imagery, physically plausible restrained motion, and a natural documentary-like tone. The audible soundscape is: Soft binocular strap and high bird calls.", | |
| "selector_query": "<Subject 2> is an older East Asian male birdwatcher with neatly cut gray hair, a brown waterproof coat, and compact binoculars. At normal speed, <Subject 2> raises his binoculars toward the open sky rather than the reeds, follows one distant movement, and lowers them without pointing it out. No intelligible dialogue is heard. The shot uses realistic cinematic imagery, physically plausible restrained motion, and a natural documentary-like tone. The audible soundscape is: Soft binocular strap and high bird calls. Identity continuity requirements: Keep <Subject 2>'s facial identity, age, neatly cut gray hair, facial proportions, skin tone, brown waterproof coat, and compact binoculars exactly consistent with his identity reference; do not redesign or substitute him.", | |
| "generation_prompt": "<Subject 2> is an older East Asian male birdwatcher with neatly cut gray hair, a brown waterproof coat, and compact binoculars. At normal speed, <Subject 2> raises his binoculars toward the open sky rather than the reeds, follows one distant movement, and lowers them without pointing it out. No intelligible dialogue is heard. The shot uses realistic cinematic imagery, physically plausible restrained motion, and a natural documentary-like tone. The audible soundscape is: Soft binocular strap and high bird calls. Identity continuity requirements: Keep <Subject 2>'s facial identity, age, neatly cut gray hair, facial proportions, skin tone, brown waterproof coat, and compact binoculars exactly consistent with his identity reference; do not redesign or substitute him.", | |
| "pictures": [ | |
| { | |
| "picture": 1, | |
| "path": "picture_01.png", | |
| "source_shot": null, | |
| "frame_index": null, | |
| "identity_anchor": "<Subject 2>", | |
| "source": "/datadrive1/v-zihanyang/evals/unityshotsbench/ltx23_joyai_echo_subject_labels_20261111/refine_assets/worker_04_tier3/subject_2_anchor_shot00_frame0060.png" | |
| }, | |
| { | |
| "picture": 2, | |
| "path": "picture_02.png", | |
| "source_shot": 5, | |
| "frame_index": 120, | |
| "identity_anchor": null, | |
| "source": "generated" | |
| }, | |
| { | |
| "picture": 3, | |
| "path": "picture_03.png", | |
| "source_shot": 7, | |
| "frame_index": 120, | |
| "identity_anchor": null, | |
| "source": "generated" | |
| } | |
| ], | |
| "audio_references": [ | |
| { | |
| "label": "<Audio 1>", | |
| "source_shot_idx": 7, | |
| "duration_sec": 5.041666666666667, | |
| "input_path": "audio_01_from_shot_07.wav" | |
| } | |
| ], | |
| "generation": { | |
| "model": "/datadrive1/v-zihanyang/Audio-Video-Benchmarks/LTX-2/models/ltx-2.3-22b-dev.safetensors", | |
| "lora": "/datadrive2/v-zihanyang/restored_training_local/ltx23/rl_checkpoint61_peft/lora_bf16.safetensors", | |
| "width": 1280, | |
| "height": 736, | |
| "num_frames": 121, | |
| "fps": 24, | |
| "num_inference_steps": 30, | |
| "seed": 50, | |
| "video_cfg_scale": 4.0, | |
| "audio_cfg_scale": 7.0, | |
| "video_stg_scale": 1.0, | |
| "audio_stg_scale": 1.0, | |
| "stg_blocks": [ | |
| 29 | |
| ], | |
| "guidance_rescale": 0.7, | |
| "video_modality_guidance_scale": 1.0, | |
| "audio_modality_guidance_scale": 1.0, | |
| "token_order": "prepend" | |
| } | |
| } | |