01 — landmark capture · 02 — crude rig · 03 — audio drive
Load a face photo, let MediaPipe find the eyes and mouth, then drive a crude jaw-drop puppet: blink on demand, or feed in a speech clip and watch the jaw open with it. MediaPipe only knows human faces — for an animal (or if detection misses), switch on Edit landmarks and drag the dots onto its eyes and mouth yourself.
Drag to hold the eyelids part-shut; 100 = fully closed.
Drives the mouth from your mic instead of the loaded clip — good for puppeteering live.
How far the mouth stretches per unit of speech volume.
Captures the puppet canvas plus whatever audio is playing, as a video file.
Doing an animal, or the auto-detect missed? Turn this on and drag the two eye dots and the mouth dot into place — drag the small square handle on each to resize its box.