Imagen Video

Imagen Video is a Google Research generative video model that creates high-definition videos from natural language prompts. Built on a cascaded diffusion architecture, it synthesizes videos at up to 1280×768 resolution and 24 frames per second from either textual descriptions or still images. The system focuses on maintaining strong semantic alignment with the prompt while preserving temporal coherence across frames, resulting in consistent motion, stable subjects, and detailed backgrounds.

Imagen Video supports various editing and conditioning modes, including text-conditioned video generation, text-driven video stylization, and generating videos from existing images. It can also render complex scenes with multiple objects, camera movements, and diverse artistic or cinematic styles.