PAINTLAB
An assistant for painting miniatures. "I have this figure and these paints. Tell me how to paint it with what I already have."
Tap a reference photo to read its colours, split the figure into zones with shadow, base and highlight, and get a paint you own, a mix of your paints or, only last, something to buy.
Built with
Next.js · React · TypeScript · Supabase · three.js · OpenAI
What it is
References give the colour. A 3D model of the figure, if you have one, gives the shape: the recesses that collect shadow and the edges that catch the light.
It answers one question, with the paints you already have on the shelf first.
How it works
Tapping a photo reads the colour you see there, and it's never presented as exact. It becomes a target you can warm up, lighten or type in by hand. Each zone of the figure has four layers (shadow, base, highlight and extreme highlight), and shadows and highlights are suggested from the base.
Every suggestion goes inventory first: a paint you own, then a mix of your paints, then a substitute, and only then something to buy. The mixer tries two, three or five paints and scores each recipe for accuracy and simplicity, always showing how far it is from the target.
Recipes turn into a painting mode for when your hands are busy: big text, a screen that stays on, and steps you move through with the keyboard or a Bluetooth page turner. AI can propose the zones of a figure, optionally.
- Next.js · React · TypeScript
- Supabase · three.js
- Colour maths in Lab, ΔE
Development
Version 0.1, end to end.
References, zones, mixes, recipes and painting mode, tested on desktop and mobile.