What problem does it solve? Writing a video-to-video prompt that changes one element of real footage without re-rolling the subject's face, performance, or camera move is difficult; this Skill produces Seedance 2.0 prompts that lock the source plate and apply only the requested transformation. ## Core Features & Use Cases - Footage Transformation Prompts: Generates prompts for adding VFX elements (fire, creatures, morphing limbs), swapping environments around a preserved subject, and relighting or regrading a shot. - Preservation Guardrails: Builds @source declarations, lock-down clauses, NON-IP specs, and 4K mode-std settings so identity, wardrobe, framing, and camera motion survive the edit. - Timed Camera Moves: Anchors crash zooms and push-ins to dialogue with dual semantic and numeric timecode anchors, including prepended-intro duration budgeting. - Use Case: A user attaches a 6-second clip of themselves talking to camera and asks to place a giant photoreal chimp on the building behind them with a zoom synced to their spoken line; the Skill outputs a plain-text Seedance prompt with the source locked, the creature integrated with contact shadows and matching light, and the zoom anchored at the measured timestamp. ## Quick Start Attach your source clip and ask for a Seedance 2.0 video-to-video prompt that adds or changes one specific element while keeping everything else identical.
