landing hero page
Results preview
Upload page
image been uploaded
analysing image
Results page
outfits page
trying this shirt on
outfits changed sucessfully

Inspiration Most people guess when they choose what to wear. They buy colors that clash with their skin's undertone. They pick fabrics that fight their skin's natural condition. Existing styling apps either ignore skin entirely or do skin analysis with no fashion follow-through. I wanted to close that gap.

What it does Aura chains two YouCam APIs into one personalized styling journey. A user uploads a selfie. YouCam Skin AI returns redness, oiliness, moisture, and skin type. Aura generates a color palette tuned to that skin profile with a clear explanation of why each color works. The user picks an outfit from a curated catalog and YouCam Cloth VTO renders the garment onto their photo. They drag a before/after slider to compare.

How we built it The frontend runs on Next.js 16 with TypeScript and Tailwind 4, deployed to Vercel. The backend runs Django 6 with Django REST Framework and Gunicorn, deployed to Render. Every YouCam call is wrapped in error handling and polling logic. The full pipeline runs as async tasks: file upload, task creation, two-second polling until success, then result rendering. Selfies go to Cloudinary first so YouCam can fetch public URLs for the VTO step.

Challenges we ran into The first challenge was timeout management. Cloth VTO can take up to 60 seconds. Gunicorn's default worker timeout is 30 seconds. The mismatch returned 502 errors on every VTO request. I raised the Gunicorn timeout to 300 seconds and reduced the polling window to fit.

The second challenge was public image hosting. YouCam VTO requires URLs for both the person and the garment. YouCam's own File API returns presigned URLs that cannot be re-fetched. I added Cloudinary as an intermediate host.

Accomplishments we're proud of The full product works end to end in production. The multi-API pipeline chains Skin Analysis and Cloth VTO in one flow. Every recommendation tells the user why. The before/after slider makes the try-on tangible. Error handling covers every YouCam error code with friendly messages. No signup, session-based flow, runs in under 60 seconds.

What we learned How to handle long-running async API tasks in Django. How to work around cross-origin image hosting requirements. How to design a multi-step product flow that keeps the user oriented.

What's next Persistent user accounts and history. A richer outfit catalog with real product links. Weather and occasion aware recommendations. Shareable look cards.

Built with: Next.js, TypeScript, Tailwind CSS, Framer Motion, GSAP, Django, Django REST Framework, Gunicorn, httpx, YouCam Skin AI, YouCam Cloth VTO, Cloudinary, Vercel, Render

Try it out links:

Live demo: https://aura-skin-stylist-ten.vercel.app