Gemini Omni is a multimodal video model for creating and editing short clips. You can begin with text, photos, or an existing video, then keep refining the result with natural language instructions.
Use it when the goal is not one random clip, but an iterative scene: preserve a subject, change the environment, adjust style, or transform source material into a new video.
What Gemini Omni does
Gemini Omni accepts different starting materials. A prompt can define the scene, images can anchor subject and style, and source video can become the base for editing.
The key difference is the iterative workflow: instead of stopping after one generation, you can continue with instructions about what should change.
Prompt-based editing
After the first result, describe what should change: background, lighting, character look, or scene style.
Up to 10 seconds
Gemini Omni Flash creates short clips with native audio for fast creative iterations.
Multiple photos
Use several input photos when a scene needs a character, object, style, or environment reference.
Where it fits
Product scenes
Turn product visuals into short videos and refine the setting or movement.
Character clips
Use photos to keep a subject or avatar direction while changing the scene.
Fast variants
Generate several short directions and keep the one that fits the edit.
How to prompt it
Write the first prompt like a short scene brief: subject, action, camera, light, mood, and format. For edits, be specific about what must stay and what must change.
