How to prompt Gemini Omni 1.1 Flash: ask for one take, and never quote the dialogue
Gemini Omni 1.1 Flash cuts a clip into several shots unless your first sentence asks for one continuous take, and it can print quoted speech on screen. The prompt shape LUVI's guide enforces in all five modes, from 1,980 credits for a ten-second clip at 720p.

Gemini Omni 1.1 Flash is Google's video model that writes picture and sound in one pass from a plain-language caption. On LUVI it runs in five modes: text-to-video, image-to-video (with an optional last frame), reference-to-video, video edit and video extend. The first three start at 1,980 credits for a ten-second clip at 720p (prices checked October 20, 2026); edit and extend are priced on the clip you upload, and the Workspace shows the estimate before you run. If you are still choosing between Omni and its neighbors, the MiniMax H3 vs Gemini Omni vs Wan 3.0 comparison covers that. This post is the how-to: the rules our prompting guide applies to Omni, and why each one exists.
Why does Omni cut my clip into several shots?
Gemini Omni 1.1 Flash cuts because that is its default, not because the prompt was unclear. Google's documentation says so directly: "By default Omni Flash will try to create a video with a few different shots." The same paragraph gives the fix: "If you need the output video to contain a single scene, you must prompt for that."
LUVI's guide turns that into a hard rule: every single-take prompt opens with a continuity clause—"In a single continuous shot", "In a single unbroken scene", "One continuous shot, no jump cuts"—in the first sentence. When the camera does something compound, the clause goes at the end as well. Two details make this more than cosmetic:
- The soundtrack is cut with the picture. Sound is generated in the same pass, so an unwanted cut also breaks the ambience or a line of dialogue.
- A camera move without the clause is the classic trigger. Ask for a reveal and the model may solve it with a cut instead of a move.
What does a good Omni prompt look like?
A good Gemini Omni prompt reads like a forty-to-sixty-word caption, not a list of tags. LUVI's guide orders it in five slots—shot framing and motion, style, lighting, location, action—then small movements in the subject or the light, then a separate sound sentence, then any exclusions.
In a single continuous shot, a slow push-in at eye level on a potter at her wheel in a sunlit ceramics studio, shallow depth of field, golden hour light through dusty windows. Her wet hands draw the clay upward into a tall cylinder while fine spray catches the light. In a voice that is calm and low, the potter says: Not yet, it needs another minute. Sound design: the hum of the wheel, wet clay under her palms, a distant radio playing a song. No text overlay on screen.
Three habits keep it working:
- Prescribe the frame, delegate the world. Say where the camera is, how it moves and what happens. Leave period detail, wardrobe logic and physics to the model.
- One narrative beat per clip. Timed beats such as "After 3 seconds, she lifts her hands" are fine inside one continuous take; separate events belong in separate generations.
- Write the prompt in English. Google states that "English (EN) is fully supported, but other languages have not been evaluated". A line of dialogue can still be in another language; the instructions around it stay in English.
How do I write dialogue, music and on-screen text?
On Gemini Omni 1.1 Flash, speech goes after a colon with no quotation marks, and quotation marks are kept for words that should appear in the picture. Google's examples put quotes only around visible text—a street sign, a number plate—and LUVI's guide draws the rule from that pattern: a quoted line risks being painted into the frame as a caption. Voice quality goes before the colon: "In a voice that is calm and low, the potter says: Not yet."
- Sound sits in its own sentence behind the label "Sound design:". Effects, ambience and dialogue all go there or next to it.
- Music has to be requested, and it works better described by its source than by its genre: "a distant radio playing a song" rather than "lo-fi hip hop".
- There is no negative prompt and no audio switch. Exclusions are short fragments at the very end: "No dialogue." "No extra sound effects." "No text overlay on screen." To remove a visible object in an edit, describe the target state—"Make the phone invisible."—rather than listing what you don't want.
Which settings actually matter?
Three settings carry most of the cost and quality on Gemini Omni 1.1 Flash: resolution, duration and aspect ratio.
- Resolution. 360p, 720p, 1080p or 4K, defaulting to 720p. Google's own reference labels 1080p and 4K as "upscaled", so 720p is the native picture; 360p is a cheaper draft tier for blocking out a shot.
- Duration. Three to ten seconds, defaulting to ten. Keep any timed beats inside the length you pick, and never write the length into the prompt—it is a setting.
- Aspect ratio. 16:9 or 9:16, nothing else. Writing "21:9" in the prose sets nothing.
One safety rule is worth knowing before you plan a story: Omni won't render photorealistic children. LUVI's guide records both outcomes—an outright refusal, or a child quietly rendered as an adult—so when a story needs one, use an illustrated or animated look, or keep the child out of frame.
The catalog still lists the older Gemini Omni Flash preview modes. At their defaults they cost more—text-to-video starts at 2,500 credits—and they have no extend mode and no last-frame slot, so the 1.1 modes are the ones to use.
How do image-to-video, edit and extend change the prompt?
Each media mode on Gemini Omni 1.1 Flash wants less description, not more, because the input already carries the picture.
- Image-to-video. Your image is the first frame. Write the motion—camera, subject, environment—and refer to people as "the subject" or "she" instead of describing them again. Add a last frame and the clip becomes an interpolation: describe the transition between the two images, not either one.
- Edit. One clause for the change, then "Keep everything else the same." Google is explicit: "Simple prompts work best for video editing. Overly descriptive prompts can lead to unintended changes." An edit takes no duration, resolution or aspect ratio; it inherits them from your clip.
- Extend. Each pass adds up to ten seconds, and Google caps the total at 40 seconds. Timestamps restart at zero where the new part begins, the continuity clause has to be repeated or the continuation may open with a cut, and the sound should be restated ("The music continues into the chorus.") or it simply carries on.
On LUVI every edit and every extension is a fresh request on the clip you upload; there is no conversation thread to continue from the previous turn. Plan the first generation to be right, and keep each edit to one change.
How we checked
- Models: Gemini Omni 1.1 Flash text-to-video, image-to-video, reference-to-video, video edit and video extend, plus the older Gemini Omni Flash preview modes for comparison.
- What was read: the live parameter contract of each mode in LUVI, the credit prices in LUVI's catalog and on the public model pages, LUVI's prompting guide for the family as the Workspace and the connector serve it, and Google's Omni documentation, on October 20, 2026. The example prompt passed LUVI's prompt check for the text-to-video mode.
- Results: from 1,980 credits for ten seconds at 720p on the 1.1 text, image and reference modes; from 2,500 credits on the older preview text-to-video mode. No outputs were generated for this post, so it makes no claim about output quality.
- Drawbacks: the colon-not-quotes rule for dialogue is a rule in LUVI's guide drawn from Google's examples; Google's page doesn't state it in those words.
In LUVI
Pick a Gemini Omni 1.1 Flash mode in the Workspace and press the feather beside the prompt: it checks and rewrites your draft against the prompting guide for the family, whose first rule is the continuity clause. Claude and ChatGPT read the same guide through the LUVI connector. What each setting does to the price is explained in how resolution and duration change the price.
New accounts start with 100 credits, and verifying a phone and a card adds up to 1,500 more. You can create a LUVI account and draft your first oner at 360p before you commit to 720p.
Sources
- Generate and edit videos with Gemini Omni Flash — Google AI for Developers, Gemini API documentation. Accessed October 20, 2026.
More from Guides
- How to prompt MiniMax H3: write a three-field document, not a captionMiniMax H3 reads a prompt as three named fields and fills any sound field you leave out with sound you didn't ask for. The document shape, the camera sentence, the dialogue tag and the defaults that catch people out, from 2,080 credits for eight seconds.
- Change one word on a finished image with GPT Image 2.5—and keep everything elseTo fix a price, date or name on a finished image, use a GPT Image 2.5 edit model: state the one change with both strings in quotes, then list what must stay. The whole frame is redrawn, so the rest looks the same but is not pixel-identical. From 49 credits an edit.
- Can Claude generate images and videos? Not on its own—here is how to do it with a connectorClaude doesn't generate photos, illustrations or video footage itself, Anthropic's help center says. Add LUVI Creator as a custom connector and Claude can quote and run 200+ image, video, audio and 3D models on your LUVI credits, with the price shown before each run.