Scene extension
Grow a clip without losing the story
The model analyzes up to 10 seconds of previous context, then adds another 10 seconds so lighting, motion, and identity stay aligned.
Released August 27, 2026
Use Gemini Omni 1.1 Flash free online for 10s scene extension to 40s, first-last frames, 4K video, and cheaper 360p drafts.
Gemini Omni Editorial ·

Context
Up to 10 seconds of prior video
Length
Extend in 10s steps to 40 seconds
Output
4K final or faster 360p drafts
New model
Google released Gemini Omni 1.1 Flash on August 27, 2026 as a multimodal video generation and editing model for developers who need stronger scene continuity, frame control, and faster draft loops.
Scene extension
The model analyzes up to 10 seconds of previous context, then adds another 10 seconds so lighting, motion, and identity stay aligned.
Frame control
Give the generator a start frame and an end frame when the shot must travel cleanly from one composition to another.
Dual quality
360p drafts can be up to 60% faster than 720p and about one third of the cost, then 4K handles the final pass.
Scene extension
Gemini Omni 1.1 Flash can read up to 10 seconds of prior video context, then grow the clip in 10-second increments until the story reaches 40 seconds. That longer memory is the core reason the model stays more consistent than short-context video tools.
It looks at as much as 10 seconds of the original video, which is far more memory than Veo’s 1-second context window.
Each extension adds another 10 seconds. Repeat the step to build a longer narrative without restarting from a blank prompt.
Keep extending until the sequence reaches 40 seconds. That range is useful for ads, explainers, and short story beats that need more than a hook.
Frame control
Specify a start frame and an end frame when a product reveal, camera move, or character action has to land on an exact composition. The model interpolates a smoother path between those two stills.
Use a first frame when the product, face, or layout must match an existing still before motion starts.
Set a last frame so the clip lands on a poster, logo, UI state, or call-to-action composition instead of drifting into a random ending.
Describe camera speed, lighting continuity, and what must not change. The model then interpolates the path between the two frames.
Quality modes
The release supports 4K high-definition output for final review, plus a fast 360p draft mode. The 360p path is up to 60% faster than 720p and costs about one third as much, so prompt tests stay cheap before a 4K render.
Choose 4K when the cut is ready for landing-page review, paid social, or any delivery that needs high-definition detail.
Use 360p to test scene extension, first-last frames, and prompt wording before spending on a 4K render.
The 360p draft mode is up to 60% faster than 720p and costs about one third as much, which keeps iteration cheap.
Workflow
01
Start from a generated shot or an uploaded video. The generator continues from where the last scene ended.
02
Write what should stay the same: subject, lighting, camera distance, wardrobe, and on-screen text from the previous 10 seconds.
03
If the beat needs a controlled transition, attach the opening and closing frames before you generate the next 10-second increment.
04
Run 360p until pacing and continuity look right, then switch to 4K for the review cut.
Use cases
This release is built for stories that need more than a 5-second hook. Extend the same scene until the action, pause, or product demo is complete.
Specify first and last frames to move from a product still into a logo end card, or from a UI screenshot into a feature callout.
The Gemini App note is practical here: continue from the last ending, or branch into a new angle without discarding the original clip.
Why this model
Google AI highlighted that scene extension uses 10 seconds of original-video context, far beyond Veo’s 1-second window. Logan Kilpatrick also noted video reference support, which helps keep characters, products, and motion language aligned.
Google AI said scene extension is based on 10 seconds of original-video context, not Veo’s 1-second window.
Logan Kilpatrick noted that the model can take video references, which helps keep motion language and identity consistent.
The release is positioned for generative video control: scene extension, frame targets, 4K delivery, and a cheaper 360p loop for prompt tests.
Announcement notes
Google AI emphasized that the new scene extension uses 10 seconds of original context, producing stronger consistency and longer narrative.
Logan Kilpatrick added video reference support, which is useful when a character or product clip already exists.
The Gemini App account said users can create or upload a video, then ask Gemini to continue from the last ending and open a new creative direction.
Related guides
Start with the broader Gemini Omni Flash workflow before moving into the 1.1 scene-extension controls.
Use a still as the first frame, then add motion and later scene extensions from that reference.
Revise camera, pacing, and end frames after a draft is already in place.
FAQ
Gemini Omni 1.1 Flash is Google’s August 27, 2026 multimodal video generation and editing model. It adds 10-second scene extension, first-last frame control, 4K output, and a faster 360p draft mode.
Google released the model on August 27, 2026 for developers who need stronger generative video control.
The model analyzes up to 10 seconds of previous video context, then extends the scene by another 10 seconds while trying to keep lighting, motion, and identity consistent.
You can keep adding 10-second increments until the sequence reaches 40 seconds. That is the announced cumulative length for scene extension.
Yes. Specify a start frame and an end frame so the generator can create a smooth transition between those two compositions.
Yes. It can output 4K high-definition video for final review after you have tested the prompt and scene extension in a cheaper draft.
360p is a fast preview path. Google said it can be up to 60% faster than 720p and about one third of the cost, which makes prompt iteration cheaper.
Google AI said scene extension uses 10 seconds of original-video context, compared with Veo’s 1-second window, which should improve consistency and longer storytelling.
Yes. Logan Kilpatrick noted video reference support, so an existing clip can guide identity and motion.
Yes. The Gemini App account said you can create or upload a video and ask Gemini to continue from where the last scene stopped, either to lengthen the story or to explore a new direction.
No. Gemini Omni 1.1 Flash is a Google model family, but this website is an independent interface and guide for a focused video workflow.
Upload or generate one shot, keep 10 seconds of context, decide whether you need first and last frames, run a 360p draft, then render 4K when the continuity looks right.
You can start Gemini Omni 1.1 Flash free online. Use 360p drafts to test scene extension first, then spend credits on a 4K render when the continuity looks right.
Gemini Omni 1.1 Flash workflow
Extend the last scene, lock the end frame, then render a 4K cut.