Point a camera
at your face.
This page calls the real Eyeconic API on every frame. No backend tricks, no canned demo loop — what you see below is exactly what your rig receives. Webcam works in Chrome and Edge. Image upload works anywhere.
What you're looking at.
Every frame is POSTed to /v1/analyze/base64, run through
MediaPipe + the Eyeconic eye model in roughly 15 ms, and returned as
the JSON payload you see in the side panel.
Frames are analysed in-memory and discarded — no pixel data is logged or stored. Schema is the same as production: 52 ARKit-compatible blendshapes, per-eye gaze vectors, head pose, emotion confidence.
Point the api field above at your own deployment if
you'd rather hit a local backend.