OpenArt adds Gemini Omni Flash for conversational video editing
By Steven Van ·
Creators can now refine OpenArt clips by chatting in plain language, with edits grounded in Gemini's real-world physics and context.
OpenArt just made conversational video editing a first-class workflow: Gemini Omni Flash is now live on the platform, letting you shape and refine clips by simply talking to them. Instead of hunting through timelines and effect panels, you describe the change you want in plain language and the model does the rest, grounded in a real understanding of how the world actually works.
Edit videos through natural conversation
The headline capability is conversational editing. Generate or import a clip, then refine it by chatting: swap a character, adjust the lighting, change the background, or restage a shot, all in everyday language. Because the edits are iterative, you can keep nudging a scene toward what you had in mind without ever leaving the conversation, then hand off to OpenART's finishing tools for upscaling and extending the result.
Grounded in real-world knowledge
What separates Gemini Omni Flash from a purely stylistic generator is that its output is grounded in Gemini's real-world knowledge. It pairs an intuitive grasp of physics, so objects move naturally and shadows behave correctly, with an understanding of history, science, and cultural context. That grounding is why edited scenes hold together instead of drifting into uncanny territory: the model reasons about how a real version of the scene would look and behave.
One cohesive scene from mixed references
The model is natively multimodal, so you can feed it a combination of images, text, video, and audio in a single prompt and have it fuse everything into one coherent scene. Reference an image for the look, a video for the motion, an audio track for the mood, and text for the direction, and Gemini Omni Flash reconciles them together, generating synchronized sound alongside the visuals rather than as a separate step. Inside OpenArt, that sits next to the platform's other top video models, editing, upscaling, and Extend Video, so a single conversation can carry a shot from idea to finished clip.
Sources: OpenArt, Google, Google DeepMind.