Google introduced Gemini Omni 1.1 Flash, an update designed to make AI video generation more controllable, faster, and more useful. The new version is aimed at developers building creative tools, audiovisual production workflows, and editing applications.
The proposal goes beyond generating a clip from a description. You can now continue scenes, define keyframes, use videos as references, and produce preliminary versions at a lower cost. What’s the goal? To make AI feel more like a production tool and less like a black box that delivers results that are difficult to adjust.
Extend scenes to tell longer stories
One of the main new features is scene extension. Starting with an existing video, Gemini Omni 1.1 Flash can continue the action from the point where it ended, while better preserving the characters’ appearance, the environment, and the narrative logic.
The model can analyze up to 10 seconds of previous context, compared with the final second used by earlier versions. This helps preserve movement, lighting, and visual continuity during generation.
Scenes can be extended in 10-second blocks until they reach a cumulative duration of 40 seconds. For a creator, this means being able to build a longer sequence without generating every fragment from scratch or constantly fixing visual jumps.
Control the beginning and end of every shot
Omni 1.1 Flash also lets you specify the first and last frame of a shot. The model generates the transition between those two points, a useful feature for planning camera movements, zooms, rotations, and continuous loops.
For example, you can tell the camera to start with a drummer in close-up and end by revealing a saxophonist next to a dancer. You can also ask the camera to move toward a television screen and connect with the opening scene without visible cuts.
This type of control is especially valuable when video generation is part of a professional process. Instead of relying only on a long instruction, the creator can establish specific visual points and let AI complete the journey between them.
360p preliminary videos for faster iteration
The new version adds 360p generation to create drafts more quickly. According to Google, these videos can be generated up to 60% faster and at approximately one-third the cost of a standard 720p output.
The resolution is enough to test ideas, review a storyboard, or compare different instructions before producing the final result. This way, you can experiment with a microscopic scene, an animation, or a camera movement without spending high-quality production resources on every attempt.
The logic is simple: first, validate the idea with a lightweight draft, and then invest in the final version. Why render something in high resolution when you still don’t know whether it works?
Professional outputs of up to 4K
When the concept is ready, Gemini Omni 1.1 Flash can generate videos in 1080p or upscale the results to 4K for professional production use.
Google shows examples that include fish captured in motion, a small animal moving through a forest, and cinematic close-ups of maple leaves illuminated by sunlight. The goal is to offer more detail, more defined textures, and an appearance suitable for finished audiovisual pieces.
The availability of these resolutions doesn’t eliminate the need to review the result. The consistency of hands, faces, objects, and movements still needs to be evaluated in every project, especially when the video will be used in advertising, entertainment, or corporate communications.
Use reference videos to preserve characters and movements
Another feature lets you include up to three seconds of video as a reference within a multimodal input. This allows the model to use movements, dance styles, and visual context to create a new scene.
In one of Google’s examples, several videos of dancers serve as a guide for animating a dog, an octopus, and a bear. Each character must perform a different type of dance, while they all appear together in the same space and in a continuous shot.
This capability opens up possibilities for editing tools, advertising campaigns, music videos, and interactive experiences. It also makes it possible to separate a character’s visual identity from the movement you want to reuse, although you’ll always need to verify that the result respects the rights and permissions associated with the reference materials.
Available for developers and subscribers
Gemini Omni 1.1 Flash is beginning to roll out across Google’s development ecosystem. Users can try it in Google AI Studio, while companies can implement it through the Gemini Enterprise Agent Platform.
Google also offers technical documentation, a cookbook, and prompting guides for integrating features such as scene extension, video references, and resolution upscaling into your own applications.
In addition, Omni 1.1 is available globally to Google AI Plus, Pro, and Ultra subscribers within Google Flow. Scene extension is also coming to those plans in the Gemini app.
The update shows where video generation is heading: less dependence on a single prompt and more tools for directing the result. AI doesn’t replace audiovisual planning, but it can speed up many stages of the process, from the first sketch to the production-ready piece.
Original source
https://blog.google/innovation-and-ai/technology/developers-tools/build-with-gemini-omni-1-1-flash