StyleAW

Inspiration

Shopping online is frustrating — you never know how something will actually look on you. I wanted to fix that by letting anyone upload a photo of themselves and virtually try on any clothing item before buying.

What it does

Fashioning is a virtual try-on app for fashion. You browse a clothing catalog, save items to your personal closet, then go to the Try-On page — upload a photo of yourself, pick a garment from your closet or upload one, and AI generates a realistic image of you wearing it in seconds.

How we built it

Frontend: Next.js + Tailwind CSS, deployed on Vercel
AI: Replicate's IDM-VTON model — state-of-the-art virtual try-on diffusion model
Database: AWS DynamoDB — stores each user's closet and try-on history keyed by guest session ID, no login required
API: Next.js API routes connecting the frontend to Replicate and DynamoDB

Challenges we ran into

Converting AVIF/HEIC phone photos to JPEG before sending to the AI model, tuning denoising steps for quality vs speed, and wiring AWS DynamoDB credentials into a serverless Vercel deployment from scratch.

Accomplishments that we're proud of

Built a working AI virtual try-on in under 3 hours — upload a real photo of yourself, pick a garment, and get a realistically generated result. Got AWS DynamoDB and Vercel fully integrated from scratch with no prior setup.

What we learned

How to integrate AWS DynamoDB into a Vercel serverless app from scratch, and how diffusion-based virtual try-on models work under the hood.

What's next for StyleAW

Real product images linked directly to e-commerce stores so users can shop the exact item they tried on, user accounts with an outfit history gallery, support for trying on full outfits at once, and a mobile app version.