Video editing without a timeline. Pick one of your videos, type what should change — make it night, slow it down, clean up the background — and Gemini Omni Flash returns the edited video. Refine it conversationally: each instruction applies to the previous result, so you talk the video into shape instead of dragging clips.
기능 스냅샷
0
timeline to learn
1
sentence per edit
keep
what you did not mention
실제 생산 워크플로우에 매핑된 실용적인 가치.
혜택
01
Describe the change, not the edit
No in/out points, no tracks, no keyframes. “Make it dusk and add falling snow” is the whole edit — the model keeps the subject, the motion and everything else you did not mention.
혜택
02
Refine like a conversation
Every edit becomes the input to the next. Too dark? Say “brighter” and the model adjusts its own result rather than starting over — your earlier changes survive each turn.
혜택
03
Your videos, one click away
The picker is your own video library — anything you generated or uploaded. Results land back in the same library, so a finished edit can be re-edited, downloaded, or posted as-is.
소스 자료에서 재사용 가능한 결과까지의 간단한 3단계 경로입니다.
단계 1
Pick a video
Choose from the videos already in your library.
단계 2
Say what should change
Type the change or tap a preset. One instruction, one edit.
단계 3
Keep refining
Each result accepts another instruction. Stop when it is right, then download.
No. There is no timeline at all — if you can describe the change in a sentence, you can make it. That is the entire interface.
They are preserved. The model edits the video you gave it and keeps the subject, motion and audio you did not ask to change — an instruction is a change request, not a re-generation from scratch.
Each turn starts from the previous result, so you can walk a different path by returning to the original video and re-describing. The original is never modified.
Gemini Omni Flash, Google's video editing model. Edits are submitted as ordinary generation tasks, so credits, progress and results behave like every other video tool on the platform.