No art here — just the truth. Every signal in the grammar, live from your camera.
Make faces, pinch, wink, nod. Screenshot the meters mid-expression and send them — that's
how the gains get tuned. Everything stays on this device.
camera: waiting for permission ·
face model: loading… ·
hand model: loading… ·
detect: – ·
video: – ·
faces seen: 0
what the tracker sees (mirrored) — landmarks drawn when detected
the palette gauge: smile warms it, brow-raise brightens it (calibrated values)