A is a shot card written in a fixed format: camera, action, emotion, timing, and the locked look of the character, product and place. It keeps every clip consistent.
With Seedance, one JSON covers a whole scene: its shots, their exact timings from the voice-over, and who says each line, so the lips match our voices.
You'll see it under each clip on the review pages. If something's off, name one line, like camera.movement: static, and only that changes.
// Kling: one shot, shortened
{
"shot_id": "H1-S01",
"duration_s": 3,
"camera": { "movement": "slow push-in" },
"action": "Emma's eyes snap open",
"emotion": "wired, exhausted",
"avoid": ["text", "logos"]
}
// Seedance: one 28 s scene, shortened
{
"part_id": "B-P01",
"duration_s": 28,
"pacing": { "shot_count": 11, "max_shot_s": 3 },
"references": { "audio_1": "B-P01.wav" },
"shots": [
{ "t": "0.0-2.6", "speaker": "EMMA",
"line": "It's 3am. Again." },
{ "t": "2.6-5.1", "speaker": "NARRATOR", … }
]
}