Build shape
AI paint simulation web product
Paint preview simulator
Paint Preview AI
A web simulator that finds paintable surfaces in a building photo and produces before/after previews.
AISimulationB2B
Product brief
The core requirement was a pipeline that could turn an upload into a before-and-after preview immediately.
Paint Preview AI lets a user upload a building photo, detect paintable surfaces, and generate before and after color previews. It is designed for moments when a team needs to agree on the outcome before any real paint work starts.
The repo shows an API pipeline for space detection, target detection, surface analysis, palette generation, and paint editing, plus a result gallery and downloads. It is an AI visualization product turned into a usable web flow.
Scope
Problem
Problem
Color decisions needed to happen before the job started, so uploads had to turn into fast, easy-to-compare previews.
Implementation
I split space detection, target detection, surface analysis, palette generation, and paint editing into an API pipeline, then added a gallery and downloads.
Best fit
This fits AI visualization products that need to show a credible preview before quoting, consulting, or construction.
Technical structure
Product evidence
Primary stack
Next.js 16 · React 19 · Tailwind
Data & sync
Gemini · Sharp · Upstash Redis · Vercel Blob
Operational surface
exterior/parking limit · gallery · usage cap
Stack
Web frontend
Next.js 16React 19TypeScriptTailwind CSSRadix UIFramer MotionZustand
Image pipeline
@google/genaisharpapp/api/*core/engines/*palette generationbefore/after processing
Storage and operations
Upstash RedisVercel Blobremote configpublic galleryusage capengine-test
Integrations
GeminiUpstash RedisVercel BlobSharp
Screens
Product evidence
Web
App
Need a similar product, internal tool, or launch-ready first version?
Send the current problem, scope, or a reference link. The next step can start from product structure and release scope.