Gemini Omni — published as OmniEditor at omnieditor.io — is a prompt-first AI video editor. The pitch is “Chat once, get a finished image or video”: you describe the result you want, drop in your own photos or clips, and an agent picks the model, writes the prompt, and iterates with you in a live session.
That agent layer is the point of difference. Rather than making you choose a model and hand-tune a prompt, OmniEditor takes an instruction like “animate this photo into a 9:16 cinematic clip” or “restyle this clip into hand-drawn anime” and handles the model selection and prompt engineering behind it. Uploads are accepted up to 100 MB.
Key Features
- Agent-driven sessions — The agent picks the model and writes the prompt, then iterates with you live.
- Text-to-video — Define subject, action, location, camera motion, style, and output format from a scene prompt.
- Image-to-video — Guide the first motion pass from a still, product shot, storyboard frame, or character reference.
- Video-to-video editing — Keep what works in an existing clip and request targeted changes in plain language.
- Reference consistency — Hold products, characters, layouts, and style references stable across variations.
- Audio-aware direction — Plan ambience, music, speech timing, lip-sync, sound effects, and beat-matched motion.
Who It’s For
Creators, marketers, education teams, and small product teams making social videos, product demos, avatar clips, explainers, short ads, streaming intros, and campaign variants.
How It Works
- Define the scene — Name the subject, action, setting, format, and the first moment the viewer should understand.
- Add creative control — Specify camera movement, shot size, lens feel, lighting, visual style, and mood.
- Lock the references — Say which product, character, layout, color, or motion details must stay consistent.
- Iterate with edits — Request focused changes after the first result rather than rewriting the whole prompt.
The site publishes a side-by-side comparison of Gemini Omni Flash against Seedance 2.0, positioning Omni Flash as the better surface for reference-led iteration and Seedance 2.0 for a polished production pass.