VIAxVOICE

Photo reenactment experimental v0.2

Experimental: live mesh-warp in your browser (~900 triangles/frame → a few fps). To make faces match, set the four alignment points — drag head / chin (top & bottom) and left / right (cheek edges) on both panels to fit each face. Head↔chin sets vertical scale, left↔right sets horizontal — so wide and narrow faces both line up.

Drop a neutral face photo and a driving clip. Fit the four points on each, then Reenact — the photo's whole face (mouth, brows, eyes) follows the clip.

1 · Face photo (neutral, front-on)
JPG / PNG — click or drop
2 · Driving clip (a performance)
MP4 / WEBM — click or drop
loading model…

Driving clip · mesh

auto-snapped to the mesh — nudge head/chin + left/right to refine on the real head-top & chin

Reenacted photo

auto-snapped to the mesh — nudge head/chin + left/right to refine on the photo's face

Scale is normalized by the head↔chin distance you set on each face (not auto guesswork), so mismatched aspect ratios line up. All 468 landmarks warp — mouth, brows, and eyes/lids. Next: drive this from VIAxVOICE audio (no clip) so a still photo speaks any line.