Gemini Omni 1.1 Flash Reference to Video From Your Images
Last verified: November 1, 2026
Gemini Omni 1.1 Flash turns the images you upload and a written prompt into a new video clip. It carries the people, products or visual style in those pictures into a scene you describe, and the output goes up to 4K. You pick a vertical or widescreen frame and a clip length. The prompt does the directing: action, setting, camera and mood.
Google documents reference-guided generation for the model. Reference images steer what appears on screen, and the text prompt describes the scene those references play out in. That split is why reference-to-video suits recurring characters, product shots and brand looks. The images carry appearance, and the prompt carries what a still picture can't, like movement, camera work and timing.
On Vidofy, your references go into the Add Image uploader and your direction goes into the Prompt box. Three choices decide how the clip is delivered: Video Duration, Resolution and Aspect Ratio. Enhance Prompt sits under the prompt box for when you only have a rough idea, and a Seed field takes a number before you generate.
Explore more reference to video
Reference-to-video specs at a glance
The inputs, clip options and output of this generator in one view.
Input
Up to 7 images (JPG, JPEG, PNG or WEBP, up to 10 MB each) + a text prompt of up to 20,000 characters
Video Duration
4, 6, 8 or 10 seconds
Resolution
360P, 720P, 1080P or 4K
Aspect Ratio
9:16, 16:9
Output
1 video per run
How do you prompt Gemini Omni 1.1 Flash with reference images?
End the prompt by naming the images as references
Give every upload one job with reference tags
Time-code the beats when several images share one clip
How do you make a video from reference images?
Five steps, from uploading references to saving the finished clip.
Upload your references in "Add Image"
Start with the images that define what has to carry over: a clean, evenly lit shot of each character or product, plus a style frame if the look matters. One subject per image is easier to direct than a busy collage, because each picture can then get a single job in the prompt. The accepted files and limits are in the FAQ below.
Direct the scene in "Prompt"
Describe the action, setting, camera movement and mood, and say which image plays which part. The "Prompt" box takes up to 20,000 characters, which is far more than one clip needs, so spend the words on direction rather than on re-describing your images. If you only have a rough idea, click "Enhance Prompt" under the box for fuller suggestions. Your prompt only changes when you pick one, so you can compare the suggestions with your own draft first.
Set "Video Duration", "Resolution" and "Aspect Ratio"
"Video Duration" offers 4, 6, 8 or 10 seconds and starts at 4. Use the longer options for prompts with several beats so each action has time to play out. "Resolution" runs from 360P through 720P, the default, to 1080P and 4K. Check how your references come through at a lower setting, then render the take you keep in 4K. "Aspect Ratio" starts at 9:16 for vertical feeds. Switch to 16:9 for widescreen edits before you generate.
Start the run with "Generate"
Check your references and settings once more, then click "Generate". "Seed" is a number field you can fill in before the run. Writing the value down next to each prompt makes your test runs easier to keep track of when you compare them later.
Download the finished video
Watch the whole clip before saving it. Check that the subjects match your references and that the motion looks clean from the first frame to the last. If it holds up, download it. If not, change the prompt rather than the images and run it again, so you can see what the change did.
Choosing between a new clip and an edit of existing footage
When to build from images here, and when the same model's editing workflow is the better starting point.
| Criterion | Gemini Omni 1.1 Flash | Alternatives | Best For |
|---|---|---|---|
| What you start from | You start from still images in the uploader plus a written prompt, and the model builds a shot around them that didn't exist before. | The Edit Video workflow of the same model starts from a clip you already have, which Google's prompt guide treats as the source video to edit or modify. | Pick reference-to-video when your material is photos, renders or style frames, and pick Edit Video when the footage already exists and only part of it should change. |
| What the result is built on | The uploaded images steer who or what appears and how it looks, while the setting, action and camera come from your prompt. | Editing works on the existing footage through plain-language instructions, which the model card describes as editing videos through conversation. | Choose reference-to-video to put a known subject into a situation you never filmed, and choose Edit Video when the original shot's timing and framing should stay the base of the result. |
Common reference-to-video problems and fixes
What usually goes wrong with clips built from references, and what to change before the next run.
Words, logos or labels in my video come out garbled
Fast or complicated movement looks warped or unnatural
My reference image won't upload
I got a single video but want to compare a few versions
Frequently Asked Questions
How many reference images can I upload, and which files work?
Add Image takes one to seven images per run, in JPG, JPEG, PNG or WEBP format. Each file can be up to 10 MB and must be at least 300×300 pixels. A small thumbnail or a tight crop saved from a web page may fall short, so export from the original instead.
Can Gemini Omni 1.1 Flash keep the same character or product across several videos?
For a series, upload the same reference images in every run and change only the action and setting in the prompt. Describe the subject the same way each time, and keep the same Aspect Ratio so framing stays comparable. Watch each clip before moving to the next shot, so you can fix a drifting detail while the prompt is still fresh.
What makes one generation cost more than another?
The credit cost of a run is worked out from the options you choose and shown above the Generate button once you're signed in. It changes with two settings, Video Duration and Resolution, so check the figure again after changing either one. Enhance Prompt suggestions don't use credits.
How do I make a video longer than one clip?
Build a longer piece from several generations, each as long as the Video Duration you choose. Plan the story as separate shots, upload the same reference images for each one, and write one shot per prompt. Then join the clips in your editor, keeping the same Aspect Ratio throughout so the cuts match.