Nano Banana Prompts: 12 Examples and a Practical Prompt Guide

Nano Banana prompts work best when they read like compact creative briefs. A useful prompt does more than name a subject. It explains what should dominate the frame, how the scene should be arranged, what visual language to use, and where the finished image will appear.

That structure matters whether you are working directly with Gemini image generation or adapting the same idea in an independent workspace such as FreeGPTBanana. The examples below are designed to be edited. Replace the subject and business goal first, then adjust composition, lighting, palette, and format one decision at a time.

If you are new to image prompting, start with the broader AI image prompt framework. For model access, current pricing, and Lite-specific limits, read the Nano Banana 2 Lite guide. This guide focuses on the naming, visual instructions, and iterative patterns that people usually mean when they search for Nano Banana prompts.

What people mean by Nano Banana prompts

Google’s current Gemini API documentation describes Nano Banana as the name for Gemini’s native image generation capabilities. At the time of review, that documentation distinguishes four models: Nano Banana 2 Lite, Nano Banana 2, Nano Banana Pro, and the legacy Nano Banana model associated with Gemini 2.5 Flash Image.

Those product names and API model IDs can change. They also do not mean that every third-party image application exposes every Gemini model or every capability. In FreeGPTBanana, use the models and controls actually shown in the Imagine workspace. Treat the prompt itself as portable creative direction rather than proof that a specific model, resolution, or editing feature is available.

In practice, a Nano Banana prompt is simply a clear instruction for generating or editing an image with Gemini’s image capabilities. It may start from text alone, combine text with reference images, or continue an existing visual through conversational edits. The same underlying discipline applies to all three workflows: say what must stay consistent and what should change.

A practical Nano Banana prompt formula

Build the prompt in five parts. You can write them as one paragraph, but keeping the decisions separate makes revision easier.

1. Subject: Name the person, object, character, product, or environment that matters most. Include essential identity, material, wardrobe, or shape details.
2. Context: Explain where the subject is, what is happening, and what visual story the image should communicate.
3. Composition: Set the viewpoint, crop, hierarchy, depth, aspect ratio, and location of any negative space.
4. Visual language: Choose a coherent medium, lighting direction, palette, texture, and level of realism.
5. Intent: State the destination, such as an ecommerce hero, profile image, campaign post, sticker sheet, or editorial explainer.

For example, “a reusable bottle on a table” leaves most important decisions open. A more useful version is:

Premium product hero image of a matte steel reusable bottle, three-quarter view on a cool gray pedestal, subtle condensation, crisp blue rim light, realistic studio reflection, wide ecommerce composition with clear negative space on the left for campaign copy.

The extra detail is not decoration. Each phrase controls a different production decision. If the output misses the goal, you can identify whether the problem is the product, angle, light, background, or intended layout.

12 Nano Banana prompt examples

These examples cover professional portraits, ecommerce images, campaigns, characters, and visual explainers. Keep the core structure, but replace details that belong to your subject or brand.

Professional portrait prompts

Professional founder portrait for a climate technology company, calm direct gaze, navy overshirt, chest-up crop, warm gray studio background, soft directional key light with natural shadow falloff, realistic skin texture, credible editorial photography for an About page.

Approachable product designer profile portrait in a bright creative studio, thoughtful expression, simple charcoal knitwear, large window light from camera left, softly blurred shelves and sketches in the background, clean neutral palette, square crop for a professional profile.

Consistent remote team headshot of a customer success lead, friendly direct gaze, warm off-white background, soft even lighting, smart-casual wardrobe in muted blue, natural skin detail, centered shoulders, square composition matching an existing team directory.

To adapt this group, keep the role and destination explicit. Change the wardrobe, expression, and background together so they support the same professional message. When likeness matters, upload a permitted reference photo and use the reference-image workflow rather than trying to describe identity from memory.

Ecommerce and product prompts

Premium skincare serum bottle standing on translucent amber glass, low three-quarter camera angle, soft sunrise gradient behind the product, controlled highlights along the label and cap, realistic glass refraction, elegant beauty campaign image with open space above for a headline.

Clean ecommerce listing image of a lightweight travel backpack, front three-quarter view, accurate stitching and zipper placement, seamless pale gray background, broad diffused studio light, crisp but natural contact shadow, product centered with consistent margin on every side.

Lifestyle product photo of a ceramic pour-over coffee set on a walnut counter, early morning window light, faint steam, linen towel and a small branch as restrained supporting props, tactile natural materials, realistic editorial photography for a specialty coffee launch.

For product work, describe details that cannot change, especially logos, closures, proportions, and materials. A reference image can help preserve product shape, but every generated result still needs a careful accuracy review before it becomes customer-facing creative.

Social and campaign prompts

Instagram launch visual for a premium matcha drink, hero can centered against layered fresh-green paper shapes, playful condensation, crisp studio highlights, energetic editorial composition, clear upper-right text-safe area, polished square campaign image without rendered copy.

LinkedIn carousel cover about practical AI productivity, confident geometric composition, cobalt and warm cream palette, one clear workflow symbol as the focal point, premium B2B editorial style, generous headline-safe space, landscape layout without final text.

Social prompts should name both the platform format and the communication goal. Ask for text-safe areas instead of relying on generated final copy. Typeset, proofread, and accessibility-check important words after generation.

Sticker and character prompts

Sticker sheet of six friendly productivity mascots, each character holding a different planning tool, consistent rounded geometry, distinct expressions, thick cream cutout border, coral, yellow, and cobalt palette, clean isolated arrangement on a pale background, no text.

Glossy 3D banana-shaped robot mascot for a creative software launch, compact friendly proportions, expressive digital eyes, yellow shell with graphite joints, three-quarter standing pose, soft studio lighting, isolated on a clean background for use in social stickers.

For a coherent sticker family, repeat the palette, border, shape language, and scale. Vary only the action, expression, or prop. This gives the set enough difference to stay useful without looking like unrelated characters.

Editorial and infographic prompts

Editorial infographic concept explaining a four-step creative workflow, four modular cards connected by one flowing line, warm white background, coral and cobalt accents, simple dimensional icons, clear left-to-right hierarchy, generous areas for final labels, landscape composition.

Magazine-style visual explainer comparing three image prompt structures, bold central diagram, restrained yellow, black, and sky-blue palette, subtle paper grain, clean grid, balanced columns with label-safe areas, contemporary editorial design without final body text.

Use generated infographics as visual direction, not automatically verified information. Add final labels and data in a design tool, then check facts, reading order, contrast, and mobile legibility.

How to adapt a Nano Banana prompt

Do not rewrite every variable at once. Start from the closest example and make a controlled sequence of changes.

First, replace the subject and intended use. A skincare hero can become a coffee-package hero while keeping the camera angle and lighting. Second, change the environment or composition. Move from a studio pedestal to a kitchen scene, or from a square feed post to a vertical Story. Third, change the visual language, such as moving from clean commercial photography to a tactile paper collage.

After each generation, describe the most important correction in plain language. Useful follow-ups include “keep the product shape and label unchanged, but soften the shadow,” “preserve the character design and change only the pose,” or “keep the layout, remove the small decorative objects, and increase the text-safe area.”

This one-variable approach makes it easier to understand which instruction produced a useful improvement. For a complete text-first workflow, see the step-by-step text-to-image guide or browse the prompt collection for additional structures.

Common prompt failure modes

Too many competing styles

Combining photorealistic, flat vector, watercolor, glossy 3D, minimalist, and maximalist instructions does not create a richer direction. It creates conflict. Choose one medium and two or three supporting visual cues.

No composition or destination

An attractive image can still be unusable if it has the wrong crop or no room for copy. State where the subject sits, which side needs negative space, and whether the result is square, vertical, or landscape.

Asking for final text too early

Gemini image models may support text rendering, but important names, prices, dates, and legal copy still require human review. Establish the visual hierarchy first. Add or verify final copy only after the overall direction works.

Treating references as permission

Only upload images you are allowed to use. A reference can communicate identity, product shape, palette, or composition, but it does not remove copyright, privacy, publicity, or brand-review responsibilities.

Changing everything in one follow-up

If a revision changes the subject, angle, background, palette, style, and format together, you lose the useful parts of the previous result. Preserve what works and name one or two corrections at a time.

Nano Banana and Gemini naming

Search results often mix consumer product names, official API model IDs, older names, and third-party aliases. Before publishing technical claims, check Google’s current Gemini image generation documentation and record the review date.

At this review, Google describes Nano Banana as a family name rather than one permanent model ID. The documentation recommends newer models for current work and identifies the Gemini 2.5 Flash Image version as legacy. That status may change, so this article does not promise that a named model will remain available or that FreeGPTBanana exposes every Google model.

FreeGPTBanana is an independent image creation workspace and is not affiliated with Google or Gemini. Use the model selector and generation settings visible in the product as the source of truth for the current FreeGPTBanana workflow.

Ready to test one of the structures above? Open Imagine with the Nano Banana prompts campaign context, then adapt the subject and intended use before generating.

FAQ

What is a good Nano Banana prompt?

A good prompt clearly states the subject, context, composition, visual language, and intended use. It preserves any details that must remain consistent while leaving less important details open to interpretation.

Is Nano Banana the same as Gemini image generation?

Google’s current documentation uses Nano Banana as the name for Gemini’s native image generation capabilities and distinguishes several models within that family. Product names and model IDs can change, so verify the current official model documentation before relying on a specific name.

Can I use a reference image with these prompts?

Yes, when the selected workflow supports references and you have permission to use the image. Explain which identity, product detail, palette, or composition should remain consistent and what should change.

Does FreeGPTBanana support every Nano Banana model?

Do not assume that it does. FreeGPTBanana supports the models and settings shown in its current Imagine interface. The existence of a model in Google documentation does not automatically mean it is enabled in every third-party product.

Can I start generating for free?

New FreeGPTBanana accounts can start with included credits. Credit use depends on the selected model, size, and generation settings. The service is account-based and does not promise unlimited or unrestricted generation.