Seedream 5.5 is now on Scenario (opens in a new tab)
Scenario
← All Models
Video

HeyGen Talking Photo

Use it ↗

Photo to talking video on a transparent background (WebM): script up to 5,000 characters in 24 voices or audio up to 5 minutes, 720p or 1080p. Best for overlays; real faces only.

Turns one photo of a person into a talking video with the background removed, delivered as WebM with an alpha channel and audio. Speech comes from a script (up to 5,000 characters, 24 library voices) or your own audio (up to 5 minutes, which overrides the script). 720p or 1080p; aspect ratios 16:9, 9:16, 4:5, 5:4, 1:1 or auto. Expressiveness low, medium or high; a motion prompt directs gestures (a nod and hand to the chest were followed). Needs a human-like face. Best for: presenters and characters layered over your own background, vertical and square explainers, stylized human characters. Not for: non-human mascots (a 3D dragon was rejected); photos without a clear face; a 3D child character was blocked by the safety filter. Instead: opaque presenter with 2K and reference video, HeyGen Avatar 6; photo plus your audio, OmniHuman 2; lip sync on existing video, HeyGen Lipsync Speed.

More models from HeyGen