Video GenerationActive
google / gemini-omni-flash
gemini-omni-flash
Generate 5s,10s 720p MP4 videos from text, or animate a still image as the opening frame. Google Gemini Omni Flash supports multi-turn, chat-based video edits.
Text to VideoImage to VideoFast Inference
Model ID
gemini-omni-flash
Provider
google
Updated
1784809005
wiro playground—google/gemini-omni-flash
Updated 1784809005
Overview
Gemini Omni Flash by Google turns a short creative brief into a 720p MP4 video clip with sound. It plans shots from your description, then renders motion, camera moves, and audio together. You can start from text alone, or anchor the clip with a still image as the first frame. It helps when you need fast video drafts, social clips, or quick visual iterations.
What you can build
- Vertical social clips for Shorts, Reels, and Stories from a single prompt
- Image-to-video animations that start from your key art or product photo
- Concept scenes for pitches, storyboards, and mood tests
- “Make it different” variations of the same idea by iterating in chat
- Simple VFX-style edits such as swapping an object or changing the setting
- Short ad-style product shots with controlled framing and lighting cues
Inputs
- A required written description of the video you want. Include subject, setting, action, camera style, and mood.
- An optional first-frame image to start the clip. Upload a JPEG, JPG, PNG, or WebP file. Leave it empty for text-to-video.
- A clip length choice of 5 seconds or 10 seconds.
- An output aspect ratio choice of 16:9 or 9:16. You can also choose auto framing, which picks 16:9.
Outputs
- An MP4 video file (video/mp4) at 720p resolution and 24 FPS.
- A short clip that matches your selected duration.
- An audio track generated to fit the visuals and pacing.
Recommended settings
- Pick 9:16 for mobile-first posts. Pick 16:9 for landscape scenes.
- Choose 5 seconds for a single action loop. Choose 10 seconds for a mini narrative.
- If you want one continuous shot, say “single continuous shot” and “no scene cuts” in your prompt.
Limitations
- Output length is limited to short clips, up to 10 seconds.
- The model may introduce multiple shots unless you ask for one scene.
- Complex motion can look unstable, especially with fast camera moves.
- On-screen text can be wrong or misspelled.
- Reference images are not copied pixel-for-pixel. The model may reinterpret details.
- Low-quality inputs reduce fidelity. Heavy compression, watermarks, or tiny subjects can cause drift.
Safety & compliance
- Content safety filters apply to both your prompt and the generated video. Enforcement can vary by region.
- All videos generated with Gemini Omni include a SynthID watermark.
- Get rights and consent for any faces, voices, logos, or copyrighted material you upload.
- Don’t generate deceptive content that misleads viewers about real people or events.
- Capabilities that can change a person’s speech exist, but Google restricts them.
Example prompts
Great starting points for gemini-omni-flash.
The raincoat figure slowly turns their head toward the camera; neon reflections ripple across the wet asphalt, steady rain falls, distant car headlights and tail lights glow through the mist, faint steam drifts up from the street. Slow push-in, continuous smooth shot.Video Generation
Fine sand streams off the dune crests in the wind, long shadows slowly shift as the sun rises higher, gentle heat shimmer ripples along the horizon. Slow aerial drift forward over the dunes, continuous smooth shot.Video Generation
The humpback whale glides slowly forward with a gentle sweep of its tail, god rays sway across its body, bubbles rise toward the surface and tiny particles drift in the current. Slow tracking motion following the whale, continuous smooth shot.Video Generation
Steam curls upward from the latte, raindrops trickle down the window glass, warm string lights flicker softly and their bokeh shimmers, autumn leaves flutter past outside. Very slow push-in toward the cup, continuous smooth shot.Video Generation
API quick start
Run gemini-omni-flash with a single API call.
POST https://api.wiro.ai/v1/Run/google/gemini-omni-flash
{
"prompt": "The figure slowly turns their head toward…",
"duration": 5,
"aspectRatio": "auto",
"inputImage": "https://your-cdn.com/input.png"
}