Made with this app
Gemini Omni Flash 1.1 is Google's multimodal video model, available here as a focused app for generating and editing short-form video. It accepts a text prompt, a still image, an optional end frame, up to ten reference images, or three reference clips, and it can edit an existing video from a plain-language instruction. Output runs from 360p to 4K at three to ten seconds, with native synchronized audio baked in. It suits creators, marketers, and developers who need precise visual continuity, readable on-screen text, and polished audio without a separate sound pass.
How to use this app
- 1
Describe what you want to create, or upload your source file.
- 2
Pick your options, then press Generate.
- 3
Watch your result appear in the gallery within moments.
- 4
Download, share, or generate again with a new idea.
What you can make
Product Close-Up with Audio
Describe a product, supply a still photo as the starting frame, and let the model animate it at 4K with synchronized ambient or voiceover-style audio. The model's world knowledge keeps labels and text on packaging legible across every frame, removing a common pain point in AI product video.
Reference-Driven Character Continuity
Upload up to ten reference images of a character or location to steer the generation. Gemini Omni Flash 1.1 holds visual identity consistent across cuts, making it practical for storyboard-style sequences, branded mascots, or recurring scene settings that must look the same shot after shot.
Editing an Existing Clip by Instruction
Drop in a source video and type a plain-language edit instruction such as 'make the sky overcast' or 'remove the background clutter.' The model applies the change without a separate compositing step, useful for quick revision rounds on footage that is otherwise finished.
Text-Overlay Motion Graphics
Because the model is grounded in Gemini's world knowledge, on-screen text renders correctly and stays sharp through motion. Use this for kinetic titles, lower-thirds, or data callouts where other generative models typically produce garbled or drifting letterforms.
Image-to-Video Story Beats
Supply a starting image and an optional end frame to define the first and last moments of a scene, then let the model interpolate the action in between. This works well for concept presentations, animatics, and pitch decks where you need controlled motion from static artwork.
Prompt ideas to try
Why creators use this app
- Native synchronized audio
- 360p to 4K Output
- 3-10s Duration
- Text, Image and Reference Input
- Reference Video Steering
- Edit a Video by Instruction
Tips for better results
Anchor Text with a Prompt
When your video needs readable words, spell them out exactly in the prompt. Gemini Omni Flash 1.1 is grounded in world knowledge, which gives it an edge on letterform accuracy, but an explicit prompt still reduces drift and ensures the model prioritizes that element.
Use the End Frame for Control
Supplying both a starting image and an end frame gives the model two anchors to interpolate between. This is the most reliable way to control arc and composition when you need a specific visual payoff at the clip's conclusion rather than open-ended motion.
Layer Reference Inputs Strategically
The model accepts up to ten reference images and three reference clips simultaneously. Use the image slots for character or object consistency and the clip slots to define motion style or camera behavior. Mixing both input types in one generation gives you finer steering than either alone.
Write Edit Instructions Precisely
When using the video-editing mode, describe what to change and what to preserve in the same instruction. A phrase like 'change the wall color to deep teal, keep the furniture and lighting identical' is far more effective than a vague directive, because it sets explicit constraints on both the edit and the untouched elements.
When to choose this app
Choose this app when your priority is 4K resolution with native synchronized audio from a single generation pass, reference-controlled visual continuity across multiple assets, readable on-screen text, or the ability to edit an existing clip through a typed instruction. Seedance 2.5 covers longer runs up to 30 seconds and accepts far more reference material, which is a better fit when duration or reference depth matters more than resolution tier. Wan 3.0 offers 30-second single-pass video with omni-reference, making it the stronger pick for extended scenes. Gemini Omni Flash 1.1 is the right choice when you need the highest output resolution of the three, or want to revise an existing clip by typing an instruction rather than regenerating it.
Frequently asked questions
Can I generate video without supplying any image at all?
Yes. The app accepts a text prompt alone. Images, end frames, reference images, and reference clips are all optional inputs. You only need to provide them when you want to steer visual continuity, define motion endpoints, or match a specific character or location to your source material.
How does native synchronized audio work in the output?
Gemini Omni Flash 1.1 generates audio as part of the same model pass that produces the video, so sound is timed to the visuals rather than added as a generic track afterward. The resulting MP4 includes the audio channel already embedded, ready for use without a separate mix step.
What is the maximum number of reference inputs I can supply?
You can provide up to ten reference images and up to three reference video clips in a single generation. These inputs steer the model toward consistent visual identity for characters, objects, locations, or camera motion style, depending on what the references depict.
Does the video-editing mode work on any uploaded clip?
The editing mode takes an existing video as a source and applies a plain-language instruction to modify it. The model interprets natural language, so you do not need to specify technical parameters. Results are most consistent when the instruction clearly states both what should change and what should remain the same.
What resolution should I choose for social or web delivery?
For most social platforms, 1080p is a practical default because it balances visual quality with file size and upload requirements. Use 4K when the destination supports it, such as a high-resolution display demo or a platform that downsamples gracefully from a higher source. 720p suits previews or drafts. 360p is useful for rapid iteration before committing to a final resolution.
How do I keep a character consistent across multiple separate generations?
Save your reference images from one generation and supply them as reference inputs in the next. The model accepts up to ten reference images per run, so you can include multiple angles or expressions of the same character to reinforce consistency. Combining image references with a descriptive prompt that names the character's distinguishing features improves coherence further.
Which AI model powers this app?
This app runs on Gemini Omni Flash 1.1, available through Arteza with no separate account or setup.
Can I use the results commercially?
Yes. Content you generate is yours to use, subject to our content licenses.
How long does a generation take?
Most generations finish in under a minute, and you can watch progress live in the gallery.