Nano Banana Examples: 20+ Prompts with Results Across 8 Categories
Nano Banana generates and edits images across dozens of styles — 3D figurines, oil paintings, professional headshots, architectural renders, face swaps, comic strips, and more. Google’s official blog showcased 10 examples when the model launched, but the community has since discovered hundreds of use cases that push it further.
This guide organizes the most effective Nano Banana examples into 8 categories with copy-paste prompts. Every prompt below works in the Gemini App or Google AI Studio. Results vary per generation, but the prompts are tested to produce consistent output when you follow the exact wording.
Pet and Animal Examples
The most viral Nano Banana trend started with pet photos. Upload a picture of your dog or cat and the model transforms it into a collectible figurine, a video game character, or a fantasy scene — while preserving the animal’s exact markings and proportions.
3D Figurine from Pet Photo
This is the example that broke the internet. Upload a pet photo and use this prompt:
“Turn this photo into a collectible figurine inside a toy box with a clear window, bold graphics, and the pet’s name on the packaging.”
The model generates a detailed 3D figurine with accurate fur colors, eye shape, and body proportions from the original photo. The toy box includes printed graphics that match the pet’s character. This single prompt category drove millions of social media shares when Google DeepMind first released the model.
Video Game Character
From Google’s official blog: a cat sitting on a sofa was transformed into a 16-bit video game character in a side-scrolling platform level with a waterfall and forest. The prompt:
“Recreate this cat as a 16-bit video game character and place it in a 2D platform video game level.”
The model maintains the cat’s color pattern and body shape while adapting it to pixel art style. Works with any animal or person — swap “cat” for your subject.
Portrait and Headshot Examples
Portrait prompts produce some of the most practically useful output. Specifying camera parameters makes the difference between a generic AI image and a result that looks photographed.
“A professional, high-resolution profile photo maintaining exact facial structure. Framed chest up. Subject in charcoal gray blazer. Soft diffused studio lighting with subtle catchlight in the eyes. Shot on 85mm f/1.8 lens, shallow depth of field, crisp focus on eyes. Clean cinematic color grading with subtle warmth.”
This prompt generates LinkedIn-ready headshots. For more artistic results:
“Golden hour portrait with tree shadows cast on the wall. Subject in a white blouse, warm highlights, soft tones. 85mm lens, f/2.2, ISO 200, 1/500s. Film look: Kodak Gold 200, nostalgic sunlight hue.”
The camera-specific parameters (lens focal length, aperture, ISO, film stock) guide the model to reproduce the exact visual characteristics of those settings. Without them, the output looks generic. With them, the result is indistinguishable from professional photography to most viewers.
| Prompt Element | Effect | Example |
|---|---|---|
| Focal length | Controls perspective compression | 85mm (flattering portraits) vs 35mm (wider, more environmental) |
| Aperture | Controls depth of field | f/1.8 (blurry background) vs f/8 (everything sharp) |
| Film stock | Controls color tone and grain | Kodak Gold 200 (warm), Fuji Pro 400H (cool), Portra 800 (neutral) |
| Lighting | Controls mood and dimension | “Soft-box studio” vs “Golden hour” vs “Film noir Chiaroscuro” |
Style Transfer Examples
Style transfer converts a photo from one visual medium to another. The model preserves composition and proportions while applying the target style.
Photo to pencil sketch. Upload any photo and prompt: “Convert this photo into a detailed pencil drawing.” The model applies realistic pencil hatching and shading while maintaining accurate proportions. Particularly effective with architectural subjects and portraits.
Photo to fantasy map. From Google’s official blog: a rocky island photo was transformed into a stylized 3D fantasy map in a brown frame on a grid. Prompt: “Transform this into a stylized 3D fantasy map in a brown frame on a grid.” Works best with island, mountain, and coastline landscapes.
Photo to different setting. Another Google blog example: a plain white house was converted to a tropical villa. Prompt: “Transform this house to have a thatched roof and bamboo columns behind a lush tropical garden.” The model keeps the house proportions while replacing materials and environment.
Photo to watercolor. Prompt: “A quiet riverside path dusted with the first snow of the season. Bare branches arch overhead while a red-scarfed fox trots along the bank, leaving small footprints in the powder. Rendered in watercolour with soft pencil outlines, muted winter tones, and a gentle, serene mood.”
Face Swap Examples
Face swap requires uploading two images in a specific order and using a structured prompt. The upload order is critical — face source first (Image 1), target body second (Image 2).
- Upload your face photo as Image 1
- Upload the target body/scene photo as Image 2
- Use this prompt: “Insert the face of the person from image one into the body of the person of image two”
- For better results, add descriptive details: “The person in Image 1 is a young woman with dark hair. The person in Image 2 is wearing a red dress at a formal event.”
The model adapts lighting, skin texture, and bone structure to match the target environment. Character consistency holds at 95% — eyes, jawline, and skin tone transfer accurately even when the target scene has completely different lighting.
For cosplay face swaps: “Make the person in Image 1 cosplay the character from Image 2, matching the outfit, makeup, and props exactly like Image 2.” The model maintains facial identity while adapting clothing, hairstyle, and styling to match the reference character.
Product Photography Examples
E-commerce and marketing teams use Nano Banana to generate product variations, lifestyle shots, and social media content from a single product photo.
The most powerful product prompt creates a full Instagram feed from one image:
“Create a 9-image Instagram feed for this product in the same aesthetic. Use different locations, angles, and compositions, incorporating people, animals, nature, and various environments while maintaining a cohesive visual style.”
This generates 9 distinct compositions with consistent product appearance across all frames. For simpler product edits: “Change the background to a clean white studio surface with soft shadows” or “Place this product on a marble countertop with morning window light.”
| Use Case | Prompt Pattern | Best For |
|---|---|---|
| White background | “Clean white studio surface, soft shadows” | E-commerce listings |
| Lifestyle context | “On a marble countertop with morning light” | Social media |
| Full Instagram feed | “9-image feed, different locations, cohesive style” | Content calendars |
| Enhancement | “Make the picture clearer” | Quick touch-ups |
Architecture and Urban Examples
Architectural visualization prompts require precise specification of materials, lighting, and render style to produce professional-quality output.
“Futuristic data centre architecture, flowing white parametric forms, neon-accented functional lighting, glass facade, integrated green zones, sustainable technology campus, cinematic corporate lighting, architectural visualisation.”
“Modern residential balcony, smart furniture integration, vertical greenery, panoramic neon city skyline at twilight, warm ambient lighting, contemporary urban lifestyle architecture, commercial render.”
“Compact modular eco-cabin, smooth rounded architecture, warm illuminated windows, integrated natural environment, sustainable hospitality design, sunset lighting, architectural visualisation.”
The key to architectural prompts is ending with “architectural visualisation” or “commercial render” — this triggers the model to produce the clean, professional aesthetic used in architecture firms rather than artistic interpretation.
Comic Strip and Creative Examples
Nano Banana generates sequential narratives with consistent characters across panels, which previously required multiple specialized tools.
Our new Google DeepMind image generation and editing model in the Gemini app opens up a whole wide world of possibility. You can keep a person or pet consistent across edits, blend multiple photos together, alter parts of an image while keeping the rest the same.
Google DeepMind — Gemini
From Google’s official blog: two cartoon character images were uploaded and the model generated a 9-part comic story featuring both characters as superheroes. The character consistency across all 9 panels maintained recognizable faces, outfits, and proportions.
For manga: “Create a 4-panel manga titled ‘I was reincarnated into another world as an AI researcher.'” The model generates panel layout, characters, speech bubbles, and sequential narrative in authentic manga style.
For humor: “Cosy modern house that looks like a 3D model error, floating yellow balconies, stairs leading to nowhere, architectural absurdity, bright sunny day, surreal comedy aesthetic.” This style of prompt consistently produces shareable creative content.
Professional and Business Examples
Beyond creative use, Nano Banana handles practical business tasks that previously required designers or specialized software.
- Infographic generation — “Look up the 5 best vacuum cleaners under $300 and create an infographic with pros and cons for each.” Nano Banana Pro produces legible text in infographics; the original model struggles with text clarity
- Wireframe creation — upload a rough sketch and prompt: “Create a clean wireframe version of this app interface with proper spacing and alignment.” Output is usable directly in design workflows
- Outpainting — extend images beyond their original borders with “Expand while preserving the pattern” or “Add a mountain in the distance.” The model generates new content that matches the existing style, lighting, and color palette
- ID photo generation — detailed prompts specifying 35x45mm at 300 DPI, neutral expression, solid light gray background produce passport-compliant output
Most Popular Nano Banana Example Categories (by social shares)
Tips for Better Prompt Results
Three elements determine whether a Nano Banana prompt produces professional or amateur-looking output.
Specify the visual style explicitly. “Photorealistic” produces camera-like output. “Watercolor with soft pencil outlines” produces art. “16-bit pixel art” produces retro game graphics. Without a style specification, the model defaults to its own interpretation.
Include camera parameters for realistic output. Lens focal length (85mm for portraits, 24mm for landscapes), aperture (f/1.8 for blur, f/8 for sharpness), ISO, and film stock (Kodak Gold 200 for warmth, Fuji Pro 400H for cool tones). These parameters map directly to visual characteristics the model has learned from training data.
Describe composition and framing. “Framed chest up” vs “full body shot.” “Low angle looking up” vs “high angle looking down.” “Centered subject” vs “rule of thirds, subject on the left.” Composition instructions prevent the model from choosing unflattering or generic framing.