← All workAI · Fashion retail

Stylvia

See yourself in the outfit — not a model.

Live previewReactTypeScriptFirebaseGemini

Overview

An AI fashion assistant that estimates body proportions from ordinary photographs and renders complete outfits onto the user's own image, bridging digital discovery and physical fit.

The problem

Online fashion returns are driven by one thing: the customer never sees the garment on their own body. Size charts are guesswork, and every product photo is a model with different proportions.

The approach

  1. Built biometric extraction that estimates 3D proportions from 2D imagery and maps them to UK, US, EUR and INT size scales.
  2. Added environmental anchoring — using doors, chairs and floors in the photograph to establish real-world scale, rather than trusting the user to input measurements.
  3. Rendered garments onto the user's actual image with fabric physics and lighting simulation across studio, daylight, evening and gala conditions.
  4. Layered AI stylist personas with genuinely distinct aesthetic logic, so recommendations have a point of view instead of averaging every taste into beige.