testltx / inputs /shot_07 /input.json
ymyy307's picture
Upload folder using huggingface_hub
2e21b00 verified
Raw History Blame Contribute Delete
3.76 kB
{
"shot_index": 7,
"raw_prompt": "<Subject 3> is a young white female audio student with auburn hair in a low ponytail, a navy rain shell, and small over-ear field headphones. At normal speed, <Subject 3> sits on the low boardwalk rest ledge and closes her eyes for several seconds, then turns her head toward a call behind the blind. 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: Headphone fabric, breeze, and a single hollow call.",
"selector_query": "<Subject 3> is a young white female audio student with auburn hair in a low ponytail, a navy rain shell, and small over-ear field headphones. At normal speed, <Subject 3> sits on the low boardwalk rest ledge and closes her eyes for several seconds, then turns her head toward a call behind the blind. 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: Headphone fabric, breeze, and a single hollow call. Identity continuity requirements: Keep <Subject 3>'s facial identity, age, auburn low ponytail, facial proportions, skin tone, navy rain shell, and small over-ear field headphones exactly consistent with her identity reference; do not redesign or substitute her.",
"generation_prompt": "<Subject 3> is a young white female audio student with auburn hair in a low ponytail, a navy rain shell, and small over-ear field headphones. At normal speed, <Subject 3> sits on the low boardwalk rest ledge and closes her eyes for several seconds, then turns her head toward a call behind the blind. 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: Headphone fabric, breeze, and a single hollow call. Identity continuity requirements: Keep <Subject 3>'s facial identity, age, auburn low ponytail, facial proportions, skin tone, navy rain shell, and small over-ear field headphones exactly consistent with her identity reference; do not redesign or substitute her.",
"pictures": [
{
"picture": 1,
"path": "picture_01.png",
"source_shot": null,
"frame_index": null,
"identity_anchor": "<Subject 3>",
"source": "/datadrive1/v-zihanyang/evals/unityshotsbench/ltx23_joyai_echo_subject_labels_20261111/refine_assets/worker_04_tier3/subject_3_anchor_shot01_frame0060.png"
},
{
"picture": 2,
"path": "picture_02.png",
"source_shot": 1,
"frame_index": 36,
"identity_anchor": null,
"source": "generated"
},
{
"picture": 3,
"path": "picture_03.png",
"source_shot": 6,
"frame_index": 120,
"identity_anchor": null,
"source": "generated"
}
],
"audio_references": [
{
"label": "<Audio 1>",
"source_shot_idx": 6,
"duration_sec": 5.041666666666667,
"input_path": "audio_01_from_shot_06.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": 49,
"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"
}
}