What problem does it solve? Vague motion instructions like "fast action" or "she feels nervous" produce weak or drifting AI video. This Skill converts body action, object physics, and choreography requests into concrete, observable motion phrases that Seedance 2.0 can actually render. ## Core Features & Use Cases - Motion Contract: Structures every action with actor, force level, timing, physical consequence, continuity requirement, and endpoint. - Physics-Forward Pattern: States mass, force, and material causes so the model computes visible consequences like landing compression, overshoot, and skid length. - Three-Beat Timing: Segments short clips into setup, action, and changed end state, optionally synced to sound beats or SFX. - Reference Motion Rules: Maps reference video to motion, camera, or blocking while excluding unauthorized likeness transfer. - Use Case: A user asks for a fight scene; the Skill rewrites it as "Character B ducks under the swinging bag, pivots left, and stops in a guarded stance" with timing beats and stability locks for hands and faces. ## Quick Start Ask the agent to write the motion and choreography for a Seedance 2.0 clip where a character slams a door and a candle flame reacts.
