A single Omni 1.2 Flash generation or edit now runs up to 15 seconds, up from 10. Scene extension adds 3 to 10 seconds per round, so one continuous story can reach 30 seconds in total without leaving the model.
Generate native 360p or 720p drafts first, compare takes quickly, then upscale only the winning result to 1080p or 4K. The draft-first workflow keeps iteration cheap while the final file stays sharp enough for client delivery.
720p output costs about $0.07 per second with Omni 1.2 Flash, down from roughly $0.10 on the previous model. A full 15-second clip at 720p costs close to what 10 seconds used to, so longer takes no longer mean bigger bills.
Use up to 6 reference images, or 3 per character when directing two characters, plus up to 3 reference videos. Products, faces, and brand details stay consistent across shots, winning 60% of Google's identity-consistency preference tests.
Keep refining the same clip instead of regenerating it. Swap a product, adjust the lighting, or change the camera angle in a follow-up instruction, and the rest of the scene stays put while the characters, framing, and audio stay locked.
Dialogue sounds more natural, lip movement matches the words more closely, and scenes with several speakers keep each voice distinct. Talking-head clips, ads with spoken lines, and interview-style scenes come out far more usable straight from the model.
Compare clip length, extension behavior, pricing, references, editing workflow, and audio quality across the two generations.
| Feature | Omni 1.1 Flash | Omni 1.2 Flash |
|---|---|---|
| Single generation length | Up to 10 seconds | Up to 15 seconds |
| Scene extension | 10-second steps, 40 seconds cumulative | 3–10 seconds per round, 30 seconds total |
| 720p price | Baseline rate | About 30% lower per second |
| Image references | Up to 3 images | Up to 6 images (3 per character for two characters) |
| Video references | Up to 3 clips | Up to 3 clips with stronger identity carry-over |
| Editing workflow | Regenerate for each change | Stateful edits on the same clip, no regeneration |
| Speech and lip sync | Native audio | More natural dialogue, clearer lip sync, multi-speaker separation |
The key clip lengths, output options, and reference limits available with Omni 1.2 Flash.
3–15s
Create clips up to 15 seconds in one pass, then extend compatible scenes by 3–10 seconds per round.
30s
Build a longer continuous scene while the model carries forward visual and audio context.
Up to 4K
Draft at 360p or 720p, then upscale a chosen result to 1080p or 4K for delivery.
6 + 3
Guide identity and style with up to 6 reference images and 3 reference videos per generation.
Three direct steps: start from text or an image, direct the scene like a short film, then refine the same clip until it is ready to deliver.
Start from a written scene or upload an image to anchor the subject and composition. Add reference images when a product or character must look the same in every shot.
Describe the subject, action, environment, camera movement, lighting, dialogue, and timing. Clips up to 15 seconds leave room for a full beat: setup, action, and payoff in one take.
Review the draft, then change the product, lighting, or camera angle with follow-up instructions instead of regenerating. When the take works, upscale it to 1080p or 4K for delivery.
Quick answers about Omni 1.2 Flash availability, length, resolution, pricing, references, editing, and audio.
Contact Omni Flash at support@omniflash.ai
Create cinematic videos with Omni 1.2 Flash. Turn text or images into 15-second scenes with consistent characters, natural dialogue, and up to 4K output.
Sample Video
