Video Generation

Gemini Omni Flash

Native-multimodal video generation with Google Gemini Omni Flash. Conversational editing Produces video with integrated audio.

Typical generation time: 1-5 minutes

Model

Using the uploaded image as a reference, create a stunning and realistic cinematic scene featuring the same person, who is joyfully interacting with pigeons in a vibrant public square. The individual, maintaining all original features—clothing, hairstyle, and facial expressions—laughs delightfully as several pigeons flutter around them, some landing gracefully on their arms, shoulders, and even their head. In this lively atmosphere, one pigeon takes flight toward a nearby tree, skillfully landing on a branch to retrieve a small morsel of food. It then returns to the person, presenting the food with a gentle gesture, eliciting a warm smile and laughter from the individual as they continue their playful engagement with the flock. The scene is filled with more pigeons soaring through the air, occasionally landing before taking off again, all contributing to the dynamic energy of the moment. Enhance the realism with natural bird movements and intricate feather animations, while a soft breeze playfully tousles the person’s hair and coat. Illuminate the scene with warm golden sunlight, employing a cinematic depth of field to draw focus on the joyous interaction. Capture smooth camera motion and ensure the image achieves DSLR-quality realism. The overall mood should be joyful and emotional, emphasizing photorealistic details. Maintain the person’s identity and original look, while integrating only the realistic interactions with the birds. Aim for ultra-detailed, 8K quality with cinematic color grading, natural lighting, lifelike motion, and highly realistic physics throughout the scene.

First Frame Image

Maximum file size: 10 MB

Use this image as the starting frame of the video. Can be combined with Reference Images.

Reference Images

Maximum file size: 10 MB

Up to 6 reference images for subject/style-driven generation. Cite them in your prompt. Cannot be combined with a single Input Image.

Video to Edit

Maximum file size: 100 MB

Edit an uploaded video: provide the clip and an edit instruction in the prompt. Not available in EEA/Switzerland/UK. Use only the video + prompt (no images).

Output aspect ratio.

Input Image

Maximum file size: 10 MB

Target length. Omni Flash has no duration parameter, so this is requested through the prompt — the model aims for it but does not guarantee it.

Output resolution at 24 FPS. Ignored when editing a video — the edit keeps the source clip’s resolution.

Elements to avoid. Folded into the prompt as "Do not include: ..." (Omni Flash has no negative-prompt parameter).

Explicitly set the generation task. Leave on Auto to detect it from your inputs.

Provide the interaction id of a previous generation to conversationally edit/refine it.

Your generated output will appear here.

Similar Models

Community