P-Video Avatar

A performance model rather than a generator: it requires an image and an audio track, and it animates the person in the image to speak the audio. The prompt is optional, because the two required inputs already specify almost everything: who is talking and what they are saying.

That makes it the cheapest route to a talking-head clip in the catalog by a wide margin. Ten and a half credits for a five-second 720p take, nineteen at 1080p. Pair it with one of the voice models and a single portrait, and a scripted piece to camera costs very little per line.

Duration is set by the audio rather than chosen: it runs as long as the track, up to fifteen seconds. That is the right behaviour for lip sync and it means the length control in the composer is not yours to set. For higher fidelity on the same job, OmniHuman 1.5 does it at 1080p with better motion and costs several times as much; for syncing to existing *footage* rather than a still, Kling Lip Sync is the one.

P-Video Avatar takes the same portrait and voice track as OmniHuman 1.5, but its prompt is optional and often best left empty, since the portrait settles the look. An expression arc written in the prompt comes back as one steady expression, so it suits high-volume takes where the acting matters less.

Read the full OmniHuman 1.5 prompting guide