Gemini Omni

Gemini Omni unveiled multimodal editing features for video that combine Gemini's reasoning and generation capabilities.

• It enables modification and fine-tuning via natural language even in the middle of the generation process.
• Examples were shown of transforming a mirror scene into liquid-like reflections, mirror-textured arms, black-and-white line art, felt dolls, and voxel art.
• Cases were also included where the motion itself of a scene is redesigned, and sound and lighting are synchronized to match scene changes.
• Google DeepMind compared this to Nano Banana for video, emphasizing a flow where generation and editing continue through conversation.