Style Me Up — Complete the Look with AI

Upload a selfie, pick a vibe, get an AI-transformed Instagram-ready look in seconds. Built on Perfect Corp APIs to reduce fashion returns and boost social commerce.

Comment

Inspiration

40% of online fashion orders get returned. The #1 reason? Shoppers can't visualize themselves wearing the item with their own style and aesthetic. They buy hoping it works — then return it when reality doesn't match the dream.

We asked: what if you could see yourself completely styled before you buy — hair, vibe, background — in under 10 seconds?

What It Does

Style Me Up is a mobile-first web app that lets shoppers upload a selfie, choose a style vibe (Day Casual, Night Out, or Power Office), and instantly receive an AI-transformed, Instagram-ready photo of themselves.

The app chains multiple Perfect Corp APIs to deliver a complete beauty transformation:

  • AI Hairstyle Generator — applies a vibe-matched hairstyle to the user's photo
  • AI Background Change — swaps the background to match the chosen scene (street, venue, office)
  • AI Photo Enhance — polishes the final result to social-media quality

The result? A downloadable 9:16 Style Card ready to share directly to Instagram Stories via the native Web Share API.

How We Built It

  1. Frontend — React app built with Lovable, deployed on Vercel. Mobile-first, works on any phone browser.
  2. Backend — Vercel serverless function handles all Perfect Corp API calls server-side (auth token, file upload, task polling).
  3. API Flow — Authenticate → Fetch real style catalog → Upload image → Run transformation task → Poll until success → Return result URL.
  4. Sharing — Web Share API opens native mobile share sheet. One tap to Instagram Stories.
  5. Performance — sessionStorage caches file_id to avoid re-uploading same photo across vibe switches.

Challenges We Ran Into

  • 64-bit integer precision — Perfect Corp style IDs exceed JavaScript's MAX_SAFE_INTEGER, causing silent corruption. Fixed by passing IDs as raw string literals via string interpolation in JSON.
  • Vite import protection — Server-side API credentials can't be imported client-side. Resolved by moving all Perfect Corp logic to a dedicated /api/transform serverless function.
  • Dynamic catalog fetching — Style group IDs must be fetched live from the API, not hardcoded. Implemented a 2-step catalog call (groups → styles) before each transformation.

Accomplishments We're Proud Of

  • Chaining 3 Perfect Corp APIs in a single seamless user flow
  • A UI clean enough to be a real product — not just a hackathon prototype
  • Solving the 64-bit JS precision bug that stumped the API integration
  • Going from zero to deployed app in under 24 hours

What We Learned

Perfect Corp's API suite is powerful but requires careful orchestration — especially around async task polling and large numeric IDs. The real insight: the best hackathon projects aren't about perfect code, they're about a clear problem, a compelling demo, and a vision judges can believe in.

What's Next

  • Full B2B Shopify plugin — brands embed "Complete the Look" on product pages
  • Makeup Transfer API integration for full beauty transformation
  • Persistent style avatar — user's look is saved across sessions
  • Analytics dashboard for brands: try-on rate, share rate, conversion uplift

Built With

Share this project:

Updates