Inspiration

Shopping often starts with a long list of products, even when someone already knows the look they want. Look Exchange starts with the shopper's style brief and brings competing outfit proposals together with personalized virtual try-on.

What it does

Shoppers browse a sample catalog or create a style brief, compare proposals from four sample stores, and preview clothing and bags on their own photo through the YouCam API. They can explore the try-on and video screens and save a favorite look. A short AI video adds another way to visualize the selected look.

The four stores and all products, prices, and sizes are demonstration samples, not real shops. Nothing can be purchased. Generated previews and videos illustrate appearance, not exact size or fit.

How we built it

The frontend is deployed on Vercel and connected to a Python FastAPI backend on Render. The backend integrates real YouCam clothing try-on, bag try-on, and video generation, with API-unit guards. The experience connects browsing, the brief, store proposals, visual previews, video, and a saved look.

Design and engineering challenges

A 5-second video took anywhere from 47 seconds to about 11 minutes. One exceeded our 300-second deadline but still completed and cost 5 API units. We extended preview/video deadlines to 180/900 seconds, retained usage reservations for timed-out tasks, and rechecked the same task to avoid duplicate generations. We also added an /example page that instantly replays a real run, so slow generation or a sleeping server would not block the demo.

Accomplishments

We brought the main screens into one deployed frontend and connected it to a deployed backend with real YouCam integrations. The result demonstrates a different starting point for shopping: describe the outcome, compare store proposals, then see the look on yourself.

What we learned

This project explores how virtual try-on can become part of comparing shopping proposals, rather than a standalone image effect. It also highlights the importance of separating visual inspiration from claims about physical fit or real inventory.

What's next

Connect permission-cleared retailer catalogs and live availability, gather shopper feedback, and improve proposal relevance and generation reliability. Real purchasing and retailer participation would be future additions.

Built With

  • fastapi
  • next.js
  • youcam-api
Share this project:

Updates

Submission history