Google I/O introduces Gemini Omni Flash for video

At Google I/O, Google introduced Gemini Omni: a family intended to create from mixed inputs, beginning with video. Its first model, Gemini Omni Flash, accepts combinations of text, images, audio and video and supports conversational editing.

Google described video generation as producing short clips, with demonstrations focused on roughly ten-second outputs. The important workflow change is iterative direction: a creator can ask for a background change, camera adjustment or other edit in natural language.

That can reduce production friction, but it does not replace creative direction or rights checks. Agencies should still validate brand accuracy, music and likeness permissions, and the provenance of uploaded source material.