运动：以深度图打斗视频作为运动参考。严格按 1:1 跟随参考——每一个姿势、节奏与运动幅度。镜头构图遵循参考视频。

Prompt I used (Seedance): MOTION: the depth-map fight video is the motion reference. Follow it strictly 1:1 — every pose, the timing, the movement amplitude. Camera framing follows the reference video. CHARACTERS: exactly as the reference. BACKGROUND: exactly as the reference. AVOID: extra people, facial distortion, twisted hands, dull gray lighting, on-screen text, subtitles, watermarks. Two-person rule: never swap the characters — each locks to ONE figure's motion track, no identity switching mid-fight.

运动：以深度图打斗视频作为运动参考。严格按 1:1 跟随参考——每一个姿势、节奏与运动幅度。镜头构图遵循参考视频。 角色：与参考完全一致。 背景：与参考完全一致。 避免：多余人物、面部扭曲、手部变形、灰暗打光、屏幕文字、字幕、水印。 双人规则：绝不互换两个角色——每个人物锁定一条独立的运动轨迹，打斗过程中不得切换身份。