Mirror Session

A quick, honest AI check of your skin and outfit before the moment that matters.

Comment

Inspiration As a medical writer working with strict regulatory documentation, I know how much confidence comes from feeling prepared before something important. Mirror Session started from a simple question: what if you could get the same kind of quick, honest readiness check for how you look before a job interview, photoshoot, or big moment — combining skin insight and outfit confidence in one place, instead of guessing in front of a mirror?

What it does Mirror Session gives users a two-part readiness check before an important moment:

  1. Upload a selfie and get an instant AI skin analysis — hydration, texture, and a simple, encouraging recommendation (not clinical jargon).
  2. Upload an outfit photo and see a virtual try-on preview, so you know how the look actually comes together. Both readings are combined into a simple, supportive summary — helping users feel prepared, not judged.

How we built it Built with Lovable (React + TypeScript), Mirror Session integrates two YouCam APIs from Perfect Corp:

  • AI Skin Analysis API for hydration, texture, and skin condition scoring
  • AI Clothes Virtual Try-On API for outfit preview Both integrations follow the same async pattern: upload the image via the File API, start the AI task, poll for completion, and parse the returned scores/composite image into the UI.

Challenges we ran into The biggest challenge was correctly parsing YouCam's nested response shapes — matching file_id and upload URLs between the File API and Task API steps, and handling headers as objects rather than arrays. Debugging runtime errors step-by-step (rather than guessing) was key to getting both the Skin Analysis and Apparel VTO flows working end-to-end.

Accomplishments that we're proud of Getting a full end-to-end AI Skin Analysis flow working — from photo upload to real, personalized skin scores displayed in the UI — as a first-time API integration built entirely through natural language prompts.

What we learned Async API workflows (upload → task creation → polling) require careful attention to exact response shapes. Reading error messages precisely, rather than guessing fixes, made debugging far more efficient.

What's next for Mirror Session Refining the Apparel VTO flow with better photo-quality guidance for users, adding a combined "coherence score" that blends skin and outfit readiness into one summary, and expanding to support multiple outfit comparisons per session.

Built With

Share this project:

Updates