Generating a short AI video is one thing; keeping multiple shots consistent so they feel like part of the same sequence is a harder challenge. Google is addressing that with Gemini Omni 1.1 Flash, its latest generative video model designed to give creators and developers greater control over what happens from the first frame to the last.
Stronger Scene Continuity Through Longer Context
The most notable update centers on how much prior footage the model can use for context when continuing a scene. Omni 1.1 can examine up to 10 seconds of previous video rather than relying only on the final second. This broader view is intended to help maintain consistency in characters, environments, and the overall direction of a scene instead of guessing what should come next.
Longer AI Videos With Room for a Beginning, Middle, and End
Creators also gain more flexibility in how long sequences can run. Videos can be extended in 10-second increments until they reach a total length of 40 seconds. While still short compared with traditional video, the added length creates more space for AI-generated clips that include a clear start, middle, and finish.
First-and-Last Frame Control and Video References
Users can supply both the opening and closing frame they want for a shot. Gemini then generates the frames needed to connect the two. This approach can support camera movements around a subject, smooth zooms between compositions, or clips designed to loop without an abrupt jump.
Video references are supported as well. Up to three seconds of existing footage can be provided to give the model extra visual context. Google indicates this should help preserve elements such as character appearance across generated scenes.
Faster, Lower-Cost Previews Before Full-Quality Output
Not every test needs to begin at maximum quality. Omni 1.1 lets developers generate 360p previews that Google states are up to 60% faster than standard 720p output and cost roughly one-third as much. These lower-resolution versions make it practical to test ideas, refine prompts, or explore multiple versions before committing more time and resources to the final render. Once the result is satisfactory, the model can output 1080p footage or upscale the completed video to as high as 4K.
Availability for Developers, Businesses, and Subscribers
Omni 1.1 Flash is already available to developers through the Gemini API in Google AI Studio. Businesses can access it via Google’s enterprise Agent Platform. The model is also rolling out globally in Google Flow for AI Plus, Pro, and Ultra subscribers. Scene extension is coming to those same subscribers inside the Gemini app, making one of the model’s key new capabilities easier to try without building a custom application.