Play a song into a video.
Drop in a song with lyrics — or any audio with a voice — and VidBnch pulls the vocal, lets you direct a storyboard, renders it with LTX-2.5, and makes the face lip-sync to that voice. No painted mouth, no face swap, open weights only.
From your song and a face, shape up to 12 black-and-white panels — the board is the contract the render inherits. Direct it, approve it. The instrument you play, not a prompt box.
Your approved board + the extracted vocal → LTX-2.5 renders the video → LSDub warps the face's own teeth, lips and tongue to the words. Out comes a clip that sings.
One render opens every 15 minutes · the line opens for judging week
Block 1 — lip-sync only. Pick a face, pick a voice clip, add the words it sings, and the Dell renders a short clip that lip-syncs to that voice. One render opens every 15 minutes.