Description

An experimental workflow demonstrating cross-modal reconstruction: a video model (MiniMax-H3) designs a motion loop, which is then analyzed by a code model (Claude Opus 5.5) to produce an identical implementation in a single HTML file using SVG and GSAP. The project shows how generative pixels can be converted back into production-ready vector animation code.

Descriptions, tags, and model credits may be AI-generated or inferred from public sources and can be incomplete or wrong. Learn more.