What problem does it solve? When asked to recreate the look and pacing of a specific reference film or website in the Vawe engine, it is easy to copy literal pixels, brand marks, or copy, which is both a taste failure and a rights problem. This Skill provides the discipline and workflow for extracting only the transferable grammar (pacing, cut rate, devices, ground tone) and rebuilding it with original content. ## Core Features & Use Cases - Machine-assisted reference study: Runs make study VIDEO=refs/ref.mp4 NAME=ref to measure shot lengths, ground patterns, motion curves, and threads, writing a committed grammar file at grammar/<name>.json. - Site reference capture: Uses make sections URL=<site> NAME=<name> plus make palette for website references, without committing captured frames. - Honesty-first authoring: Requires naming upfront what cannot be matched (licensed fonts, captured products) and scaffolding a replaceable skeleton via make scaffold TYPE=recreation. - Use Case: Given a Framer-style hero video, study its grammar, then author an 11-second 16:9 scene where a prompt types, an artefact builds beside it with no hard cut, and the same artefact is later shown from three viewports, using invented generic content. ## Quick Start Study the reference film at refs/ref.mp4 and then scaffold a recreation scene JSON that follows its measured grammar with original generic content.
