The idea
Two or three phones capture a live, rotatable body mirror for checking a haircut, clothes, posture, or a movement.
How it works
- Place phones around a person and align their views in a shared coordinate frame.
- Reconstruct a live body representation from the camera coverage, updating as the person turns or moves.
- Pinch and rotate the view like a game character to inspect areas a single mirror cannot show.
AI’s role: Track the person and label requested areas such as hair, clothing, or posture; reconstruction must combine the actual camera views and handle occlusion.
First demonstration
Place two phones in front and behind a person and render a rotatable body view. Check whether the rear view updates smoothly during a slow turn.
What to solve next
Can the live model remain responsive and recognizable as hair, loose clothes, and limbs move across camera views?