Virtual try-on and product swaps
Upload a clip in reference mode and add the garment or product image. Ask for one swap, for example: Replace the jacket with the one in the reference image; keep her movement and the camera unchanged.
15-Second ClipsEdit Your VideoNative Speech & SoundUp to 4K
Gemini Omni 1.2 Flash is Google's video generation and editing model for clips of up to 15 seconds with speech, music and sound effects. Create from text, frames or up to 6 reference images, or upload a video and change its background, lighting or products.
Google built 1.2 for longer takes, closer reference following and edits to finished clips. These are the jobs it handles best in the studio, with a prompt pattern for each.
Upload a clip in reference mode and add the garment or product image. Ask for one swap, for example: Replace the jacket with the one in the reference image; keep her movement and the camera unchanged.
Choose 12 or 15 seconds so the hook, the demo and the end card fit in one generation. Write the beats in order (first, then, finally) and put any on-screen text in quotes.
Upload a clip of 2 to 15 seconds and name one change: the background, the lighting, the weather or the season. The result keeps your clip's length, and you pay only for its seconds.
Put each line in quotes and say who speaks it, such as: The woman on the left says "Did you bring the tickets?" Speech, music and sound effects are generated with the picture.
Upload an opening and a closing image to fix where the shot starts and where it lands. This mode runs 4 to 10 seconds; related compositions give the cleanest transition.
Describe one unbroken move, such as an FPV flythrough, an orbit or a slow pull-back, and add "no cuts". Google reports fewer jump cuts on long moves in 1.2.
Google's recommended workflow: draft cheap, lock the references, fix details with edits, and render the final take once.
Test the prompt and timing at 720P, where a 10-second clip costs 50 credits. Keep the resolution low until the motion and the sound are right.
Use clear images of the face, the outfit or the product, up to 6. Say in the prompt what each image is for, so the details hold from the first frame to the last.
Upload your best take in reference mode and change one element per request. Asking for several changes at once tends to alter parts you wanted to keep.
Run the approved prompt again at 1080P (6 credits per second) or 4K (10 credits per second). Check faces, hands, on-screen text and sound before you publish.
Compare the same output settings before choosing a tier.
| Setting | Gemini Omni 1.2 FlashThis page | Gemini Omni 1.1 Flash |
|---|---|---|
| Example cost (site credits) | 40 | 64 |
| Output resolution | 720P / 1080P / 4K | 720P / 1080P / 4K |
| Text / frame duration | 4 / 5 / 6 / 8 / 10 / 12 / 15s | 4 / 6 / 8 / 10s |
| First & last frames | Supported | Supported |
| Material-image references | Up to 6 images | Up to 7 images |
| Video references | Up to 1 clip · 15s total | Up to 1 clip · 10s total |
| Generated audio | Included · no off switch | Included · no off switch |
Reference-video inputs can change the total. Check the generator’s quote for your request.
Generate AI images and videos with subscription credits or credit packs. They share one credit balance; image costs vary by model and resolution, while video costs vary by model and duration. New users get 10 free trial credits.
Cancel anytime
For individuals & casual creators
Ideal for personal projects & starters
For professional creators & teams
Most popular choice for professional creators
Designed for studios & enterprise scale
Enterprise solution for highest productivity
It is Google's video generation and editing model, released on November 5, 2026. It makes clips with speech, music and sound effects from text, images or a video, and follows 1.1 Flash with longer clips and closer reference following.
5 credits per second at 720P, 6 at 1080P and 10 at 4K, so a 10-second 720P clip costs 50 credits. Editing an uploaded video is billed for the seconds of that video. The studio shows the exact quote before you generate.
4, 5, 6, 8, 10, 12 or 15 seconds. First and last frame clips are limited to 10 seconds, and an edited video keeps the length of the clip you upload (2 to 15 seconds).
Yes. Choose the reference mode, upload one clip of 2 to 15 seconds and describe the change. You can add up to 5 reference images to the same request, for example a product to swap in. Without a video, you can use up to 6 images.
No. Audio is generated with every clip from your prompt, and there is no audio upload or switch to turn it off. Replace the track in a video editor if you need your own voice.
Every output carries Google's invisible SynthID watermark; there is no visible logo. You can use what you create commercially, subject to the NanoBananas Terms of Service.
Use 1.2 for clips longer than 8 seconds, for editing a video you already have, or to save credits: 8 seconds at 720P costs 40 credits on 1.2 and 48 on Veo 3.1 Fast. Use Veo 3.1 Quality for polished 8-second shots where fine detail matters most.
Only 16:9 and 9:16 are available. Fast sports, complex physics and very dense prompts can still go wrong, and Google says it is still improving them. Check faces, hands and on-screen text before you publish.
Audio is included automatically. This model does not offer an audio toggle here.