Veed Fabric 1.0

Veed Fabric 1.0 is Veed's talking-video model. It takes a single still image and an audio file, and generates a video of the subject speaking the audio with synced lip, head, body, and hand motion.

Most lip-sync models start from existing video and re-sync the mouth to new audio. Fabric starts from a single still image and generates the entire talking video from scratch, useful when you don't have source footage of the character at all.

Realistic photos, illustrations, 3D renders, anime, clay-style mascots, and other stylized art. The same model handles a portrait or a cartoon character.

Any language. Fabric is driven by the input audio, swap the voice track and the lip-sync follows the new language, useful for content localization from one source image.

Up to 5 minutes per generation, well beyond the typical 8–20 second ceiling for talking-avatar models, long enough to cover full explainers, ads, and short tutorials in one run.

Open Copilot, attach a character image and a voice track, and pick Veed Fabric. For talking-head social content, ads, or multilingual explainers where you only have a still and an audio file, Fabric is the right pick.