XNews

Gemini Omni Flash 1.1 Video Extension Update

Gemini Omni Flash 1.1 Video Extension Update

Google DeepMind and Pika Labs announced the release of Gemini Omni Flash 1.1, an update to the 'anything in, anything out' world model. The new version introduces video extension capabilities, allowing clips to be lengthened in 10-second increments with up to 10 seconds of video context. It supports first and end frame references, up to three reference videos, and output resolutions up to 4K. The update also includes 360p draft previews and 4K upsamplers. The feature set is designed to give users more control over AI-generated video by specifying start and end points and using multiple reference videos to guide the output.

Key insights

The Update Focuses on Granular Control and Continuity in AI Video Generation

Gemini Omni Flash 1.1 introduces features that give users precise control over video generation boundaries. The ability to specify first and end frames, along with up to three reference videos, allows for directed video extension rather than open-ended generation. The model supports extending clips in 10-second increments with up to 10 seconds of context, enabling the creation of longer, coherent sequences. The output pipeline now scales from 360p drafts for quick iteration to 4K final renders, optimizing the workflow from prototyping to production.

Source posts