Gemini Omni Video

Native lip-syncing and audio generated in the same pass as the picture is the main idea behind Gemini Omni Video.

The AI video generator produces 1080p clips from a text prompt or from a reference image, so you can start from words or from a still.

Gemini Omni Video processes text, image, audio and video tokens in a single sequence, which means sound does not have to be added afterward in a separate editing step.

The platform describes this as the source of its coherence, realistic physics and millisecond-accurate alignment between speech and mouth movement.

It is built on what it calls Google's unified multimodal Transformer architecture.

That matters most if you make short talking-head clips, cinematic scenes with dialogue, or image-to-video animations and want the audio to arrive finished.

Categories on the site cover text to video, image to video, short video generation and lip sync.

Those are the tasks to test it on first, since claims about quality are the platform's own and best checked with your own prompts.