MirrorProof

Inspiration;

anytime i ahop online i often receive clothes that look different from product photos. MirrorProof was created to provide visual evidence before buying.

What it does;

MirrorProof generates a virtual try-on using YouCam, then compares the preview with a real garment photo. It evaluates colour, pattern, silhouette, length, and visible garment preservation to produce a reliability score. MirrorProof measures visual consistency—not physical fit, comfort, or sizing.

How we built it;

We used Next.js, React, TypeScript, YouCam’s API, Sharp for image analysis, Supabase, and Vercel. The score is calculated as: [ R = 0.30C + 0.25P + 0.20S + 0.15L + 0.10I ]where each variable represents a visible comparison metric.

Challenges;

The main challenges were handling inconsistent image quality, integrating the YouCam workflow, protecting users’ photos, and clearly communicating what the score can and cannot prove.

What we learned;

We learned that responsible AI products need transparent measurements, privacy-conscious processing, and honest limitations not just convincing generated images.

What’s next;

We plan to improve the computer-vision model, support more garment categories, and help retailers build trustworthy reliability records from verified comparisons.

Built With

• github
• next.js
• playwright
• postgresql
• react
• sharp
• supabase
• tailwind
• typescript
• vercel
• vitest
• youcam