Google has introduced Gemini Omni 1.1 Flashan updated multimodal model focused on giving creators greater control over AI-generated video. One of its biggest additions is the ability to extend an existing video scene rather than generating every clip from scratch.
According to Google, the model can analyse up to 10 seconds of preceding footage when extending a scene. This significant improvement over the earlier approach, which relied on only the final second of a video. The additional context is designed to help maintain elements such as characters, lighting, movement and the overall narrative as the video continues.
This means creators can generate a clip and then ask Gemini to continue the story from where the previous scene ended, opening up more possibilities for longer AI-generated sequences.
AI Videos Can Be Extended Up To 40 Seconds
Gemini Omni 1.1 Flash can extend videos in 10-second increments, with a cumulative extension of up to 40 seconds. Rather than generating a 40-second video in a single pass, the longer sequence is created by repeatedly continuing an existing clip.
The feature could be particularly useful for filmmakers, content creators and developers experimenting with AI storytelling. A short generated scene can be developed into a longer sequence, while creators can also take the story in different creative directions.
Google’s announcement demonstrates the feature with a futuristic car travelling across a reflective road in the sky, showing how a generated scene can continue beyond its original endpoint.
More Control With First and Last Frames
The update also gives creators greater control over how a video begins and ends. Developers can provide a first frame and a final frame, allowing Gemini Omni 1.1 Flash to generate the footage that connects the two images.
This can be useful for creating camera movements, transitions, zooms, and looping sequences. Instead of relying entirely on a text prompt to determine how a shot develops, creators can define important visual points and let the model generate the movement between them.
This approach could make AI video generation more useful for projects that require greater visual control and consistency.
360p Drafts Make Experimentation Faster
Gemini Omni 1.1 Flash also introduces a 360p draft mode for creators who want to test ideas before producing higher-resolution footage. Google says 360p drafts can be generated up to 60% faster than 720p versions and cost around one-third as much.
The feature is aimed at rapid experimentation. Creators can try different prompts, camera movements, and story ideas at a lower resolution before selecting their preferred result for a higher-quality generation.
This could make the process of developing AI videos more efficient, particularly when multiple versions need to be tested.
Gemini Omni 1.1 Flash Supports Up to 4K Upscaling
Once a video has been finalized, Gemini Omni 1.1 Flash can produce higher-resolution output, including 1080p and 4K. Google positions the capability for more polished and production-oriented video workflows.
The model also accepts video references of up to three seconds, allowing creators to provide additional visual information when generating new footage. This can help carry elements such as appearance, movements, and visual style into the generated scene.
Availability of Gemini Omni 1.1 Flash
Gemini Omni 1.1 Flash is available to developers through the Gemini API in Google AI Studio, while enterprise users can access it through the Gemini Enterprise Agent Platform. Google has also made scene extension available to AI Plus, Pro, and Ultra subscribers through the Gemini app, with the model rolling out through Google Flow as well.
Conclusion
Gemini Omni 1.1 Flash marks another step toward more flexible AI video creation. With scene extension, frame-to-frame generation, faster 360p drafts, and upscaling up to 4K, Google is giving creators more control over both the creative and production process. The ability to continue an existing scene could be especially valuable for building longer and more consistent AI-generated stories.