Fitdrop
An interactive, draggable exploration of fashion from 1980 to 2025, generated with AI image pipelines and matter.js physics.
Fitdrop is a personal, interactive exploration of fashion history from 1980 to 2025, and an experiment in the speed of AI-assisted prototyping. Outfits are generated from a JSON dataset of looks through a Gemini image pipeline using reference-pose subject injection, then dropped into a draggable matter.js physics playground. The repository is structured as a walkthrough so others can vibe-code their own version.
1 Review
Log in to leave a review.
Tried Fitdrop over a weekend build sprint to see how matter.js handles draggable fashion cards. Physics feel decent once loaded, but initial asset load was heavier than expected, had to lazy-load the AI-generated images to stop the tab from choking on mobile. No auth or backend needed which is refreshing for a portfolio piece. Decade filtering works cleanly. Worth studying the collision setup if you're building similar drag interactions.