Google's New AI Video Makes Clips Longer and Easier to Control

Google just gave AI video creators something they have always needed: control. On Aug. 27, the company announced Gemini Omni 1.1 Flash, an update to its AI video generator that lets people extend clips and guide how scenes move. Instead of forcing the AI to invent a whole shot, you can now tell it where to start and where to end, and it will fill in the motion between those two points.

The old version of Omni could only make clips of 3 to 10 seconds, and trying to make longer videos often led to characters or settings changing between shots. The new version fixes that. It can extend a scene in 10-second steps, up to a total of 40 seconds, using the previous 10 seconds of footage as context. That extra information helps the AI keep faces, clothes, and locations looking the same as the video continues.

There are still limits. The model can only generate up to 10 seconds in a single pass, so longer videos are built piece by piece. It also can only reference very short inputs, and the 1080p or 4K output is upscaled rather than created natively. But Google added cheaper previews: 360p video costs just $0.03 per second, while 720p is $0.10 per second. That makes it affordable to test many versions of a scene before buying the high-quality final version.

What this means for creators is less starting over. AI video has been great for impressive one-off clips, but not for practical, editable work. Omni 1.1 moves it closer to a real production tool, where you can iterate, refine, and keep the details consistent. It is not yet a way to generate a full-length film, but it is a meaningful step toward AI video that acts like a professional collaborator.