Start building with Nano Banana 2 Lite and Gemini Omni Flash

Google DeepMind’s June 2026 release describes an image-to-video workflow: generate a still image with Nano Banana 2 Lite, then supply it to Gemini Omni Flash as a reference for animation. The first model handles image generation; the second handles video generation and conversational editing. Developers can connect them through the Gemini API, as illustrated by the company’s product-video demo. The announcement also gives launch limits that matter when building this workflow: Omni generated ten-second clips, and its API did not yet support audio references or scene extension. It explicitly warned that accepted video references were not being processed correctly.