Assign one role to each input
Name the image, video, or audio reference and say exactly what to preserve from it. Separate subject identity, appearance, motion, environment, and sound so references do not compete.
Gemini Omni Prompt Builder
Free first-viewport prompt builder
Gemini Omni Prompt Builder for Multimodal Video
Turn a video idea and its text, image, audio, or video references into one structured Gemini Omni brief. Define what each input controls, sequence the edit over time, then copy the prompt for an official Gemini workflow.
This page builds prompts only. NanoPic does not currently run Gemini Omni or upload your references from this page. The separate Veo 3.1 link opens NanoPic's own generator and must not be interpreted as a Gemini Omni request.
Structured Gemini Omni brief
Creative goal: Create a restrained eight-second product film in which an unbranded cobalt ceramic vase moves from a quiet studio still life to a finished gallery display. Input roles: Use the first image only for the vase shape and glaze color. Use the short reference video only for the slow turntable rhythm. Use the audio clip only for room tone; do not copy any speaker identity. Timeline and edit plan: 0-2s: locked medium-wide studio frame. 2-6s: the vase turns slowly as light travels across the glaze. 6-8s: extend into a clean gallery setting and hold the final front three-quarter view. Camera: One smooth, low-amplitude push-in; no cut, orbit, zoom pulse, or sudden reframing. Audio: Soft studio room tone, one ceramic contact sound, no dialogue, no music. Continuity and exclusions: Keep vase geometry, glaze marks, scale, and orientation stable. No readable text, logo, watermark, extra object, duplicate vase, or black border. Requested output: 16:9, 1080p
Copying is free and starts no provider task. Verify the model code, supported inputs, safety rules, and output settings in the official interface before submitting.
Plan the parts Gemini Omni needs to understand
Google documents Gemini Omni as a multimodal video model for generation, extension, and conversational editing. A useful brief makes every input's job explicit instead of asking the model to infer which reference controls identity, motion, sound, or style.
Write change over time
Describe the opening frame, timed actions, edit point, and landing frame. For extension or editing, list the source elements that must remain unchanged before describing the new material.
Constrain continuity
Lock subject count, geometry, clothing, product shape, camera direction, dialogue ownership, and forbidden artifacts. Review one major change at a time in a multi-turn workflow.
A reviewable Gemini Omni workflow
Choose the task before the references
Decide whether the request is generation, extension, first-and-last-frame planning, or editing. Do not combine unrelated tasks in the first turn.
Map every reference
For each file, state its role and what must not be copied. Confirm you have rights to the people, audio, brands, and source footage you provide.
Copy and verify in the official interface
Paste the structured brief into an official Gemini workflow, select the available Gemini Omni model and supported settings, then review the visible provider terms before submitting.
Refine with one controlled change
Inspect continuity, timing, audio, text, hands, geometry, and borders. Ask for one targeted edit while explicitly preserving the parts that already work.
Gemini Omni prompt builder FAQ
Does this NanoPic page run Gemini Omni?
No. It is a free prompt-planning tool and sends no generation request. Use Google's official Gemini interface for a Gemini Omni task. NanoPic's Veo 3.1 link is a separate generator with a different model.
Which official model code should I check?
Google's current model documentation lists Gemini Omni Flash as gemini-omni-1.1-flash. Model names, availability, supported inputs, and deprecation dates can change, so verify the official model page when you submit.
Can Gemini Omni use more than text?
Google describes text, image, audio, and video inputs, plus video generation and editing workflows. Exact input combinations and limits depend on the current API and interface.
Why separate input roles?
A role statement reduces ambiguity. For example, one image can control subject appearance while a video controls motion and an audio clip supplies ambience, without implying that every reference should be copied in full.
Does copying a prompt cost NanoPic credits?
No. Copying and editing the brief on this page are local browser actions. Opening another generator does not submit a task; any available task cost must be shown in that generator before submission.