16 · Live
Capture Assistant
A green frame and one hint at a time until the shot is right — then a scan that works first time.
How it works
Live read
Every 400 ms on the camera feed: brightness, sharpness, the person's extent and centring. A head-to-toe guide turns on when the frame is clean; otherwise one hint: more light, hold still, step back, move left.
One check before the scan
On capture or upload, a vision model confirms one person, full body, facing camera, arms clear of the torso, fitted clothing, face clear, nothing obstructing.
Block or warn
Hard problems (no one / more than one person, not full body, obstructed) require a retake; soft ones warn and let you use it anyway. The same checks run on the portrait.
Real output · sample data
On the sample likeness both passes clear; a loose dress or feet out of frame would have stopped the scan before it spent a generation. Fictional model and brands.
Interactive demo
Technical specs
- Live checks
- Light · blur · head · feet · distance · centring
- Model checks
- People · pose · arms · clothing · face · obstruction
- Cost to you
- Included — counted as an analysis