Faces

Faces is a local-first facial-vocabulary trainer. It isolates one structural feature at a time, tests retrieval with a session-scaled Leitner deck, and turns the same vocabulary into a top-to-bottom field card.

What it includes

• Field card — choose one term for each of 11 structural surfaces and copy a concise description.
• Drill — study randomized two-view exemplars, hide the label, and answer a four-choice probe.
• Structural atlas — browse all 62 terms by facial surface.
• Appearance reference — optionally record the nearest Monk Skin Tone (MST) swatch without treating tone as anatomy, race, or sun-response type.
• Session metrics — inspect descriptive recall, viewing time, and answer latency. These are learning aids, not a validated assessment.

Progress stays in the browser. There are no accounts, analytics, database writes, or network requests for learner data.

Development

Requirements: Node.js 22 and npm.

npm ci
npm run dev

The app builder preview contract uses port 8080. The production build targets Vercel through the existing TanStack Start/Nitro configuration.

Verification

npm run lint
npm run typecheck
npm test
npm run build

npm run check runs all four gates. The unit suite covers catalog integrity, feature isolation, retrieval choices, Leitner scheduling, MST data, and saved state migration in addition to the app-builder template contracts.

Saved-state compatibility

The current schema is version 2 under faces-trainer. On first load it falls back to the retired lineament-trainer key, migrates either legacy cards or newer deck progress, maps the old structural skin entry to appearance.skinToneId, removes retired skin cards, and keeps the last 400 valid trials.

Monk Skin Tone attribution

This project reproduces the ten official colour swatches from:

Monk, Ellis. “Monk Skin Tone Scale,” 2019. https://skintone.google

The Monk Skin Tone Scale is licensed under CC BY 4.0. A swatch is a close visual reference only: skin appearance changes with lighting, cameras, screens, and the sampled area. The scale should not be used as a proxy for race or for Fitzpatrick phototype.

See THIRD_PARTY_NOTICES.md for the attribution record.

License

Project code is available under the MIT License. Third-party content retains its own license.