Gemini Omni Flash 1.1 Reference to Video is Google’s multimodal video-generation endpoint designed to create coherent video by combining multiple reference images, short reference videos, and text instructions. Rather than relying on a single starting image, it reasons across the provided references to build a unified video while maintaining important subject characteristics such as appearance, clothing, and voice. The endpoint supports up to three reference videos, with each video limited to three seconds, alongside reference images. Generated videos can be produced at 360p, 720p, 1080p, or 4K, with 16:9 or 9:16 aspect ratios.