F.A.C.T
An augmented reality fabric try-on built with three others from UBC’s Aerial Robotics and Rocketry Club. We used a webcam and body tracking to let someone preview a pattern on themselves.
01
Making a swatch behave like clothing
I built F.A.C.T with three other members of UBC’s Aerial Robotics and Rocketry Club. We wanted to preview fabric patterns on a person through a webcam. The challenge was mapping a flat pattern to the wearer as they moved, with controls for the pattern’s scale, tiling, and position.
02
Tracking and mapping
MediaPipe estimates body pose, then OpenCV uses that information to position the fabric overlay. Tracking supplies a reference for where the pattern belongs; the mapping step determines how much of the pattern appears and how it repeats across the garment area.
03
Keeping the preview interactive
Flask and SocketIO connect the video-processing pipeline to the interface. The wardrobe view exposes the pattern controls, so the user can adjust the appearance while looking at the preview rather than repeatedly generating separate images.
In pictures
