What problem does it solve? Multi-scene AI videos often drift in visual style between scenes, fall out of sync with the voiceover, and force full re-renders when a single scene fails quality checks. This Skill imposes a production discipline that keeps visuals consistent, timing locked to narration, and every scene independently regenerable. ## Core Features & Use Cases - VO-first timing: Write and measure narration duration per beat before generating any visuals, so the voiceover acts as the master clock for all motion and images. - Style-anchor keyframe: Generate one approved anchor frame that locks palette, texture, and framing, then reuse it across every scene to prevent style drift. - Resumable per-beat state: Store script.json, vo_durations.json, and per-beat assets so a failed scene can be regenerated alone without rebuilding the whole video. - Use Case: Planning a 12-scene product reel where each scene's image prompt, motion, and caption must match a fixed look and align exactly with a Vietnamese voiceover track. ## Quick Start Ask the assistant to plan a multi-scene video from your script using VO-first beat timing, a style-anchor keyframe, and per-beat resumable state files.
