Hold the radio. Follow the take.
Still + audio. Lips and body follow the sound.
The look

Cover still is card art for the Night-ride radio take first-click — mid-shot Volt courier holding a pocket radio on a rain-slick rooftop ledge. Sample InfiniteTalk video QC pending — no paid infinitetalk / NanoGPT run this PR. Job is still + audio → InfiniteTalk speaking/singing video (lips + body), not the talking-avatar graph (Muse → MiniMax Speech → LongCat). Open the graph from Examples — Muse paints the still; keep the Speech bootstrap or swap in a take under ~15 s; Avatar / lipsync stays on infinitetalk at people=single / 480p. Inspect body motion and lip timing. This model follows the sound; it does not invent a TTS script.
Cover still is card art for the Night-ride radio take first-click — mid-shot Volt courier holding a pocket radio on a rain-slick rooftop ledge. Sample InfiniteTalk video QC pending — no paid infinitetalk / NanoGPT run this PR. Job is still + audio → InfiniteTalk speaking/singing video (lips + body), not the talking-avatar graph (Muse → MiniMax Speech → LongCat). Open the graph from Examples — Muse paints the still; keep the Speech bootstrap or swap in a take under ~15 s; Avatar / lipsync stays on infinitetalk at people=single / 480p. Inspect body motion and lip timing. This model follows the sound; it does not invent a TTS script.
Turn a still and an audio take into a speaking/singing courier clip where lips and body follow the sound — not a TTS→LongCat intro. First-click is a Volt charcoal/cyan night-ride radio take: mid-shot courier, pocket radio chest-high, rain-slick rooftop. Distinct from Night-courier spoken intro (Muse → MiniMax Speech → LongCat).
Make it yours
- Open the graph — save first if the canvas already has your work. Undo brings yours back.
- Rewrite Still brief, the Radio take (bootstrap) line, and Delivery, or swap Speech for an uploaded clip under ~15 seconds. Leave Avatar / lipsync on
infinitetalkat single / 480p. - Cover still is card art for the mid-shot courier + pocket radio — not a paid InfiniteTalk QC clip. Sample video QC pending. Open the graph and hit Run with your key. Inspect body motion and lip timing. This model follows the sound; it does not invent a TTS script.
- Listed from about $0.09 for InfiniteTalk (duration/resolution) plus Muse ~$0.01. Speech only if you keep the bootstrap. Prices and results vary. This page has no reviewed first-party InfiniteTalk clip yet.
Download workflow · Original sampled graph · Explore MCP tools · awesome-noodles