Nano Banana, Gemini Omni and Flow for campaign creative

#01Which Google creative tool does what?

Google calls Gemini Omni its "next-generation video generation and editing model". Earlier Google video work was branded Veo. The current Gemini API pricing page lists Omni for video, so we use that name.

#02What do the API prices look like?

From the Gemini API pricing page, paid tier, 5 November 2026:

In the Gemini app and Flow you pay through your plan instead. Paid Google AI plans include Flow credits and, on Plus and above, Omni Flash video. Check the current plan page for the credit amounts. Do not quote a number from memory.

#03How do you write a good image prompt?

Use a shot brief, not a sentence.

Subject: [product or person], exactly as in the attached reference image.
Scene: [setting, time of day, props].
Composition: [close-up / mid / wide], [angle], subject on the [left third], space for headline on the right.
Style: [photographic, natural light, 35mm look] or [flat illustration, brand colours #0B3D91 and #F2A900].
Aspect ratio: [1:1 / 4:5 / 9:16 / 16:9].
Do not: change the logo, add text, add extra products, add hands holding the product.

Then edit in follow-ups instead of regenerating:

• "Keep everything, but change the season to winter."
• "Make the laptop screen show the attached dashboard screenshot."
• "Make three variants with different backgrounds, same product and lighting."

#04How do you make short video with Omni and Flow?

1. Start from a storyboard. Ask Gemini to turn your script into 4-6 shots with duration, framing and action.
2. Make keyframes first. Generate a still for each shot with Nano Banana and get it approved.
3. Animate in Flow or the app. Use the approved still as the starting frame, and describe camera movement and action.
4. Keep shots short. Several short shots edit better than one long generation.
5. Edit in your normal tool. Add logo, legal lines, captions and music in your editor, not in the generation.

#05What rules should a brand set?

Our view: AI creative is best as a speed tool for exploration and variants. Final hero assets for a launch still deserve a designer's eye, because small errors in hands, text or product details are what audiences notice.

#06What about Asset Studio in Google Ads?

At Google Marketing Live 2026, Google announced Asset Studio as part of its ad creative tools. If you run Google Ads, generating variants inside the ad account keeps assets next to their performance data. Use the Gemini app or Flow for concept work, and Asset Studio for production variants that need to live in campaigns. See Ask Advisor for the agent that can help set those campaigns up.

#07Worked example: Ledgerline

Ledgerline is a fictional accounts-payable software company. Details are illustrative.

Ledgerline needed creative for a LinkedIn campaign about faster invoice approvals, with a small budget and no photo shoot.

Walkthrough

1. The designer wrote shot briefs and attached the real dashboard screenshot.
2. She generated the hero, then edited instead of regenerating: lighting, desk props, screen content.
3. Legal checked that no generated person was presented as a real customer.
4. She animated the approved still in Flow, in three short shots, and replaced one shot where the UI text warped.
5. She added logo, captions and the legal line in the editor, then stored prompts and outputs together in Drive.

#08Common mistakes

• No reference image. The model invents your product.
• Generating text, prices or logos. Add them in post.
• One long video generation. Short shots are easier to fix.
• Fake testimonials with generated faces.
• Ignoring credit budgets. Video uses credits fast.

#09Try it yourself

Take one approved campaign image. Write a shot brief with a reference, and generate three variants in new aspect ratios using follow-up edits. Then animate one variant into a 5-8 second clip. Note how many edit turns each took and what went wrong. That's your realistic production estimate.

#10Quick quiz

1. What's the approximate API price for Gemini Omni Flash video?

Answer: Whatever the Gemini API pricing page lists that day. Do not memorise a rate.

2. Why upload a reference image?

Answer: So the model keeps your real product or UI rather than inventing one.

3. What should you add in post, not in generation?

Answer: Logos, legal lines, prices and captions.

#11What should you check before the next lesson?

• You have a shot brief template and a reference image library.
• You've written brand rules for AI creative.
• You know your plan's Flow credits.

Previous: Gemini Notebook (formerly NotebookLM) for marketing research. Next: The Gemini API, AI Studio and Agent Platform. Or return to the Gemini for Marketers track. Related: ChatGPT image generation for campaigns, Responsive search ad assets, Gemini plans and pricing.

#12Common questions

What is Nano Banana? Nano Banana is Google's name for its Gemini image generation and editing models. The newest in the API is Nano Banana 2.1, an update to Nano Banana 2 (Gemini 3.1 Flash Image).

What is Gemini Omni? Gemini Omni is Google's video generation and editing model, announced at Google I/O 2026. Check the Gemini API pricing page for the current Omni Flash video price before you quote it.

Are Gemini images watermarked? Google adds an invisible SynthID watermark to images and video made with its generative models. Check your ad platform rules on labelling AI-generated content.