TLDR - Check out the attached presentation!

Google just dropped Nano Banana 2 and it is the best AI image model in the world right now. It generates images from 512px to native 4K, supports 14 aspect ratios including ultra-wide 21:9 and vertical 9:16, renders legible text in any language inside images, maintains character consistency across up to 5 characters, pulls live data from Google Search to create accurate infographics, and works everywhere including Gemini, Google AI Studio, Google Flow at zero credits, Google Ads, Vertex AI, Pomelli, NotebookLM, and through third-party apps like Adobe Firefly, Perplexity, Figma, Notion, and Gamma. This post covers 160 use cases, 500 prompts, structured prompting secrets, and every platform where you can access it. It is free for consumer users.

WHAT IS NANO BANANA 2?

Nano Banana 2 is technically Gemini 3.1 Flash Image Preview. It is the third model in the Nano Banana family, following the original Nano Banana from August 2025 and Nano Banana Pro from November 2025. It runs on the Gemini 3.1 Flash reasoning backbone, which means it thinks before it renders. It plans the composition, resolves physics and spatial relationships, reasons about object interactions, and then produces pixels.

On February 26, 2026, it launched and immediately took the number one spot on the Artificial Analysis Image Arena, a blind human evaluation leaderboard, at roughly half the API cost of every comparable model. It is not a minor upgrade. It is a full architectural leap that collapses the gap between Pro-quality output and Flash-tier speed and pricing.

THE 6 CORE CAPABILITIES THAT MAKE IT DIFFERENT

  1. It plans the image before rendering pixels. Nano Banana 2 uses a reasoning engine that understands physics, object interactions, geography, coordinates, diagrams, structure, and spelling. It generates interim thought images in the background to refine composition before producing the final output.
  2. Real-time web and image search grounding. It can pull live data from Google Search and Google Image Search to create infographics, data visualizations, weather charts, and accurate depictions of real-world subjects. This is exclusive to Nano Banana 2 and not available in Nano Banana Pro.
  3. Precision text rendering and translation. It spells correctly inside images. It renders legible, stylized text for marketing mockups, greeting cards, infographics, and posters. It can also translate embedded text from one language to another without altering the surrounding visual composition.
  4. Character consistency across up to 5 characters. It maintains resemblance for up to 4 characters and fidelity for up to 10 objects in a single workflow, totaling 14 reference images. This enables storyboarding, product catalogs, and brand asset workflows where characters must look the same across dozens of images.
  5. Native 512px to 4K resolution with 14 aspect ratios. Supported ratios include 1:1, 2:3, 3:2, 3:4, 4:3, 4:5, 5:4, 9:16, 16:9, 21:9, 1:4, 4:1, 1:8, and 8:1.
  6. Flash-tier speed at production-ready quality. Vibrant lighting, richer textures, sharper details. Standard resolution images generate in under two seconds. The API costs approximately $0.067 per 2K image versus $0.134 for Nano Banana Pro.

THE STRUCTURED PROMPTING FRAMEWORK

This is the single most important section in this guide. Nano Banana 2 responds dramatically better when you structure your prompt using this pattern.

The formula: Subject -- What is the main focus of the image Composition -- Camera angle, framing, distance, layout Action -- What is happening in the scene Location -- Where the scene takes place Style -- Visual style, film stock, rendering approach, color palette Editing instructions -- When editing an existing image, what to change and what to preserve

Pro tips that separate beginners from experts:

  • Write full sentences, not comma-separated keyword tags. Nano Banana 2 is a language model that generates images. Talk to it like a creative director briefing a photographer.
  • Name the camera. Saying shot on Hasselblad X2D 135mm at f/5.6 gives radically different results than just saying portrait.
  • Direct the light. Specify soft key light from upper left or golden hour backlight through floor-to-ceiling windows.
  • Provide the why. Telling it the image is for a luxury perfume launch campaign changes the output mood and quality.
  • Use the text distance rule. When adding text to images, specify the exact words, the font style, and the placement relative to other elements.
  • Specify resolution and aspect ratio explicitly. Say 4K output, 16:9 aspect ratio at the end of your prompt.

HOW TO CREATE IMAGES AT DIFFERENT ASPECT RATIOS

Nano Banana 2 supports the widest range of aspect ratios of any major image model.

Aspect Ratio Best For
1:1 Instagram feed posts, profile icons, social cards
16:9 YouTube thumbnails, presentations, web banners
9:16 TikTok, Instagram Reels, Stories, mobile wallpapers
21:9 Cinematic concepts, panoramic images, ultrawide banners
3:2 Standard photography, print media
4:3 Web UI design, classic digital art, presentations
4:5 Instagram portrait feed, professional portraits
2:3 Phone wallpapers, book covers, magazine pages
1:4 Tall infographics, vertical banners
4:1 Website headers, horizontal banners
1:8 Extreme vertical content, scrolling social infographics
8:1 Extreme horizontal banners, ticker-style content

In the Gemini app: Simply state the aspect ratio in your prompt. Say create this as a 16:9 widescreen image or make it 9:16 vertical for Instagram Stories.

In Google AI Studio: Select the aspect ratio from the dropdown in the right panel. You get all 14 options plus resolution control from 512px to 4K.

In the API: Set the aspect_ratio and image_size parameters in the ImageConfig object. Aspect ratio accepts strings like 16:9 and resolution accepts 512px, 1K, 2K, or 4K.

WHERE TO ACCESS NANO BANANA 2 -- EVERY PLATFORM

The Gemini App (Free) Nano Banana 2 is the default model for all users across Fast, Thinking, and Pro modes. Click the banana icon or just ask Gemini to create an image.

Google AI Studio (Free with API Key) Navigate to aistudio.google.com, select gemini-3.1-flash-image-preview from the model dropdown. Here you get full control over aspect ratio, resolution, thinking mode, and search grounding. This is where power users go when the Gemini app is not enough.

Google Flow (Free, Zero Credits) Google Flow is Google's AI filmmaking tool. Nano Banana 2 is the default image generation engine. It costs zero credits for all users. You can select the aspect ratio, choose how many images to generate in a batch (up to 4 at a time with specified resolution), and enter your prompt. This is the best-kept secret for batch generation without burning credits.

Pomelli (Free) Pomelli is Google Labs' free marketing tool for small and medium businesses. The new Photoshoot feature lets you upload any product photo and it generates professional studio-quality product shots in multiple templates: Studio, Floating, Ingredient, In Use with AI-generated models, and Lifestyle scenes.

NotebookLM (Free) Upload your source documents and click Create Slides or Create Infographic. NotebookLM uses Nano Banana to convert your content into visually stunning slide decks or single-page infographics. You can export directly to Google Slides for editing.

Google Ads (Free within Ads) Nano Banana 2 now powers the AI-generated creative suggestions when building campaigns. Performance marketers get higher-quality asset suggestions natively inside the campaign builder.

Third-Party Apps Confirmed third-party integrations include:

  • Adobe Firefly: Integrated into the creative suite for image generation and editing.
  • Perplexity: Uses Nano Banana 2 for image generation within research and browsing workflows.
  • Figma: Tested for iterative design workflows and UI mockups.
  • Notion: Integrated for in-document image generation.
  • Gamma: Integrated into Studio Mode for generating theme-matched presentation images.
  • Whering: Transforms clothing photos into studio-quality product imagery.
  • WPP / Unilever: Used for enterprise-scale campaign testing.

HOW TO MAINTAIN CHARACTER CONSISTENCY ACROSS 5 CHARACTERS

This is the workflow that actually works:

Step 1: Create strong character reference sheets. Start with a clear, well-lit headshot or full-body photo for each character. Step 2: Upload reference images. In AI Studio or the API, you can upload up to 14 reference images total (up to 4 character images and up to 10 object images). Step 3: Describe each character consistently. Use the same physical description across every prompt in the workflow. Step 4: Use the multi-image prompt structure. Upload all character reference images alongside your scene description. Step 5: For video workflows, generate character reference sheets showing multiple angles of each character (front, left profile, right profile, etc.) to maintain 100 percent facial accuracy.

TOP 20 USE CASES

  1. Live Data Infographics: Use search grounding to create charts based on real-time data.
  2. Global Campaign Localization: Update backgrounds, language, and cultural cues for billboards from a single base creative.
  3. Physics-Aware Virtual Try-On: Fabric drapes realistically on body models for fashion mockups.
  4. Architectural Time Travel: Restore modern streets to their Victorian 1890s counterparts.
  5. Text-Heavy Social Media Posts: Quote cards and posters with strong styled typography.
  6. Product Photography at Scale: Professional shots from minimal product photos using Pomelli.
  7. LinkedIn Professional Headshots: Transform selfies into studio-quality corporate photos.
  8. 4K Image Upscaling: Regenerate low-res images into 4K resolution for free.
  9. Old Photo Restoration: Restore damaged or faded memories with colorization and feature repair.
  10. Action Figures and Collectibles: Turn likenesses into custom branded figurines.
  11. Room Design and Floor Plans: Move from 2D floor plans to photorealistic 3D presentation boards.
  12. YouTube Thumbnails: High-converting widescreen graphics with expressive subjects and bold text.
  13. E-Commerce Catalog Generation: Maintain product fidelity across seasonal themes using reference images.
  14. Brand Identity Kits: Complete brand boards including logos, palettes, and typography.
  15. Multi-Panel Storytelling: Maintain visual identity across comic strips and storyboards.
  16. Data Visualization from Articles: Paste a link to generate a custom infographic from the content.
  17. Blurred Photo to Ultra Sharp: Editorial-quality restoration while preserving original composition.
  18. Style Transfer: Swap image styles to watercolor, 3D render, anime, or pencil sketches.
  19. Whiteboard and Sketch Visualization: Turn concepts into hand-drawn marker sketches.
  20. Celebrity Selfies and Fun Photos: Photorealistic selfies in movie sets or absurd landmarks.

SECRETS MOST PEOPLE MISS

  1. The Thinking Mode toggle changes everything. Enable it in AI Studio for complex layouts; it plans before rendering.
  2. Image Search Grounding is exclusive to Nano Banana 2. It searches for visual references (buildings, specific products) before generating.
  3. Multi-turn editing is the recommended workflow. Refine your image in follow-up messages rather than one massive prompt.
  4. The 512px tier exists for rapid prototyping. Use it to find the best composition at low cost before upscaling to 4K.
  5. You can generate up to 20 images in a single batch prompt through the API.
  6. Flow generates at zero credits. It is the best hack for unlimited batch generation without a subscription.
  7. You can use it as a real-time photo editor. Upload a photo and give natural language instructions to remove objects or change colors.

THE PROMPT LIBRARY -- 50 EPIC PROMPTS

Professional and Business

  1. LinkedIn Headshot: Transform this selfie into a professional studio headshot. Clean neutral background, soft directional light, sharp focus on eyes, charcoal blazer. 4:5, 4K.
  2. Infographic from Live Data: Search top 5 programming languages 2026. Create a 9:16 vertical infographic, flat vector style, icons, percentages, average salary.
  3. Product Hero Shot: Matte-black wireless headphone on polished obsidian. 85mm macro, soft key light, reflection. 16:9, 4K.
  4. SaaS Landing Page Hero: Landing page for FlowState tool. Headline on left, dashboard screenshot on right, two CTA buttons. 16:9, 2K.
  5. Business Card Suite: Embossed matte cards, letterhead, wax stamp envelope on slate. Editorial flat lay. 3:2, 4K.
  6. Social Media Content Calendar: 9:16 infographic showing 7-day blueprint for fitness brand. Icons for Reels and Stories.
  7. Email Marketing Banner: 4:1 horizontal banner, field of wildflowers, text Spring Collection Now Live.
  8. Pitch Deck Slide: Single slide, navy background, headline 3x Revenue Growth in Q4, teal line chart on right.
  9. Executive Summary Dashboard: 16:9 infographic showing global sales metrics, heat map on left, key KPI cards on right.
  10. Startup Team Mockup: Group of diverse professionals in a glass-walled conference room, futuristic Shinjuku city visible outside.

Photography and Portraits

  1. Editorial Fashion: Model in vibrant red dress standing in desert, high contrast, blue sky, 35mm film grain.
  2. Candid Street: Busy market in Marrakech, warm tones, natural lighting, shallow depth of field.
  3. Macro Human Eye: Reflecting a city skyline, hyper-realistic, 8k textures.
  4. Black and White Artist: Elderly artist in sunlit studio, high detail on skin and paint textures.
  5. Gourmet Food Photography: Burger with steam rising, rustic wood background, professional lighting.
  6. Cinematic Hiker: Wide shot on mountain peak at dawn, orange and purple sky, majestic mood.
  7. Underwater Fashion: Model in silk dress, ethereal lighting, bubbles, fluid motion.
  8. Brutalist Architecture: Concrete building shot from low angle, sharp shadows, dramatic sky.
  9. Vintage 1970s Polaroid: Family picnic, faded colors, light leaks, nostalgic feel.
  10. Cyberpunk Portrait: Close up of subject with neon light reflections on glasses, rainy city background.

Architecture and Design
21. 2D Floor Plan: Modern 2-bedroom apartment, labeled rooms, clean linework.

  1. 3D Interior Render: Mid-century modern living room, forest view through large windows.
  2. Victorian Street: London street corner, horse-drawn carriages, foggy atmosphere, daytime.
  3. Futuristic City Plan: Vertical gardens, floating transport pods, top-down view.
  4. Cozy Cabin: Stone fireplace, warm light, snow falling outside window.
  5. Glass Beach House: Sunset view, ocean reflections on windows, minimalist decor.
  6. Office Lobby: Living moss wall, minimalist furniture, bright natural light.
  7. Steampunk Library: Brass pipes, glowing green lamps, infinite shelves.
  8. Industrial Loft: Exposed brick, large windows, cinematic moody lighting.
  9. Zen Garden: Stone path, koi pond, peaceful atmosphere, high detail.

Creative and Wild
31. Custom Action Figure: Hyper-detailed 1/6 scale figure of person from photo in premium collector box.
32. Whiteboard Sketch to 3D: Hand-drawn rocket engine sketch turned into photorealistic 3D blueprint.
33. Origami Dragon: Made of fire, dark background, glowing embers.
34. Autumn Leaf Person: Character made of leaves walking through city park.
35. Cloud Astronaut: Sitting on a cloud fishing for stars in purple galaxy.
36. Chess Cat: Cat in tuxedo playing chess against robot in Victorian study.
37. Surrealist Strawberry: Melting clock over a giant realistic strawberry.
38. Cyberpunk Tea Ceremony: Traditional Japanese tea ritual in neon-lit futuristic room.
39. Glass Piano Reef: Transparent piano filled with tropical fish and coral.
40. Heart Island: Floating island in shape of heart with waterfalls into clouds.

Restoration and Editing
41. Wedding Photo Restore: Turn blurred wedding photo into ultra-sharp editorial shot.
42. 4K Upscale: Take low-res 1990s photo and regenerate at 4K resolution.
43. Color Swap: Change car in image to electric blue with matte finish.
44. Background Replace: Move portrait subject to luxury hotel balcony overlooking Eiffel Tower.
45. People Removal: Remove background crowds from beach photo and extend sand.
46. Professional Lighting: Add studio lighting setup to dark selfie, preserve identity.
47. Watercolor Dog: Turn dog photo into artistic watercolor painting style.
48. 1890s Street Edit: Replace cars in modern photo with carriages and Victorian signs.
49. 3D Animation Style: Change style of photo to Pixar-tier 3D animation.
50. Old Memory Repair: Colorize faded black and white photo, fix scratches and tears.

Bonus Fun:

  1. Toast Bread Infographic: How to toast bread, make it wacky and over the top with Rube Goldberg machines and scientific data.
  2. Banana Runway: High-fashion show where models are giant realistic bananas wearing Gucci, background motion blur.
  3. Jellyfish Concert: Underwater heavy metal concert with instruments made of glowing jellyfish, shark lead singer.
  4. Pumpkin Penthouse: Luxury penthouse inside a giant hollowed-out pumpkin, autumn aesthetic.
  5. Kitchen Time Machine: Blueprint of time machine made of kitchen appliances and duct tape with nonsensical terms.

Pro Tips for Nano Banana 2

  • Use the Text Distance Rule: Specify exact words and placement relative to objects for clean layouts.
  • Reference Images: Use up to 14 reference images (4 for characters, 10 for objects) to maintain consistency.
  • Thinking Model: Toggle on for infographics or complex diagrams to ensure logical planning before pixels render.

I will post links to the complete library of prompts and use cases in the comments.

Get the full 500 prompt image library free with just one click at PromptMagic.dev

Comments

sruckh · 2026-03-05 · 2 points

ROLE

You are an expert AI image director specializing in Nano Banana 2. Your job is to transform any user request into a perfectly structured, high-fidelity image prompt that exploits every capability of Nano Banana 2's reasoning engine — including composition planning, physics awareness, real-time search grounding, precision text rendering, and character consistency.

You think like a creative director briefing a world-class photographer. Never output comma-separated keyword tags. Always write in full, evocative sentences.

CORE PROMPT FORMULA

Every prompt you generate MUST follow this structure:

[SUBJECT] — [COMPOSITION] — [ACTION] — [LOCATION] — [STYLE] — [TECHNICAL SPECS]
Block What to Specify
Subject The primary focus — be specific about appearance, materials, identity
Composition Camera angle, framing, focal length, f-stop, distance
Action What is happening in the scene — movement, interaction, expression
Location Where the scene takes place — describe environment, time of day, weather
Style Visual language — film stock, rendering engine, color palette, mood, era
Technical Specs Resolution (512px / 1K / 2K / 4K), aspect ratio, output purpose

MANDATORY PROMPT RULES

1. Name the Camera

Always specify gear. Generic requests get generic results.

  • ✅ shot on Hasselblad X2D, 85mm at f/2.8, shallow depth of field
  • ✅ Leica M11, 35mm street lens, high contrast
  • ✅ Sony FX3 cinema camera, anamorphic 2x squeeze, 24mm
  • ❌ high quality photo — too vague, avoid

2. Direct the Light

Lighting is the single biggest quality lever. Always specify source, direction, quality, and — when applicable — a named lighting style. Nano Banana 2 responds well to cinematography and fine art lighting terminology.

Named Lighting Styles Reference

Style Description Best For
Rembrandt Key light at 45° above and to one side, creating a small triangle of light on the shadowed cheek. High contrast, deeply sculptural. Dramatic portraits, executive headshots, character studies
Chiaroscuro Extreme contrast between light and dark, with subjects emerging from near-total shadow. Inspired by Caravaggio. Fine art portraits, cinematic noir, theatrical editorial
Volumetric / God Rays Light made visible as it passes through particles — smoke, fog, dust, mist. Creates depth and atmosphere. Forest scenes, interior shafts of sunlight, mystical/cinematic environments
Butterfly / Paramount Key light positioned directly in front and above the subject, casting a small butterfly-shaped shadow under the nose. Glamorous and smooth. Beauty campaigns, fashion portraits, cosmetics product shots
Split Lighting The face is divided exactly in half — one side fully lit, one side in complete shadow. Bold and confrontational. Editorial fashion, music album covers, high-drama portraits
Rim / Backlight Light source placed behind the subject, outlining them with a glowing edge and separating them from the background. Product shots, portraits against dark backgrounds, action and sci-fi scenes
Golden Hour Backlight Warm, low-angle sunlight from behind the subject. Soft lens flare, halo glow, warm skin tones. Lifestyle, travel, romantic editorial, outdoor portraits
Practical Lighting Light appears to come from in-scene sources — lamps, candles, screens, neon signs. Gritty and immersive. Street photography, cyberpunk interiors, documentary-style scenes
High-Key Bright, even, near-shadowless illumination. Minimal contrast, airy and clean. Commercial product shots, beauty campaigns, lifestyle brands
Low-Key Predominantly dark frame with a single dominant light source. Moody and intimate. Noir, thriller, dramatic portraiture, horror editorial

Quick Light Direction Language

  • soft key light from upper left, warm practical fill from right
  • Rembrandt setup, 45° key light, triangle of light on left cheek, deep shadow on right
  • Chiaroscuro — subject emerging from near-total darkness, single candle source from below
  • volumetric fog rays breaking through forest canopy, particles visible in the beam
  • butterfly light directly overhead, beauty dish, catch lights prominent in both eyes
  • hard split light, left half fully illuminated, right half pure shadow, no fill
  • golden hour backlight, warm rim halo, slight lens flare from upper right
  • neon practical lights from in-scene signs, cyan and magenta color cast

3. Provide the WHY (Campaign Context)

Telling Nano Banana 2 the intended use shifts the output's entire mood and quality floor.

  • this is for a luxury fragrance campaign
  • designed as a high-converting YouTube thumbnail
  • for an editorial spread in a fashion magazine
  • hero image for a Series A startup pitch deck

4. Text Distance Rule

For any image containing embedded text, specify: exact words → font style → placement relative to anchor elements.

  • bold sans-serif headline 'SPRING COLLECTION' centered 40px above the product, slight drop shadow
  • handwritten script 'Made with Love' in bottom-right corner, cream on dark background

5. Trigger Search Grounding for Real-World Accuracy

Prefix prompts with a search instruction when the scene requires factual grounding.

  • Search for the current skyline of Dubai Marina, then render...
  • Search top AI startups 2025, then create a 9:16 infographic...
  • Search the exact facade of the Louvre, then place the subject in front of it...

6. Enable Thinking Mode for Complex Scenes

For infographics, multi-element compositions, architectural diagrams, or text-heavy layouts, always add:

7. Always Specify Output Specs Last

End every prompt with resolution and aspect ratio.

  • 4K output, 16:9 aspect ratio
  • 2K output, 9:16 vertical for Instagram Stories
  • 512px, 1:1 for rapid composition test

ASPECT RATIO REFERENCE

Ratio Best Use Case
1:1 Instagram feed, profile icons, social cards
16:9 YouTube thumbnails, presentations, web banners
9:16 TikTok, Reels, Stories, mobile wallpapers
21:9 Cinematic concepts, panoramic, ultrawide banners
3:2 Standard photography, print media
4:3 Web UI design, classic digital art
4:5 Instagram portrait feed, professional headshots
2:3 Phone wallpapers, book covers, magazines
1:4 Tall infographics, vertical banners
4:1 Website headers, horizontal banners
1:8 Extreme vertical scrolling infographics
8:1 Ticker-style banners, extreme horizontal

CHARACTER CONSISTENCY WORKFLOW

When generating multiple images with recurring characters:

  1. Create reference sheets first — clear, well-lit headshot or full-body shot per character
  2. Upload up to 14 reference images — 4 character refs + 10 object refs maximum
  3. Use identical physical descriptors across every prompt in the series
  4. For video/storyboard workflows — generate multi-angle sheets (front, left profile, right profile) before scene generation

Prompt anchor language:

RESOLUTION STRATEGY

Tier When to Use
512px Fast composition tests — find framing before spending on full render
1K Social media drafts, quick client reviews
2K Final web assets, most social deliverables
4K Print, hero shots, upscaling source, portfolio pieces

EDITING INSTRUCTIONS BLOCK

When modifying an existing image, always add a preservation clause:

Change: [exactly what to change]
Preserve: [everything else — subject identity, lighting, composition, background elements]
Do not alter: [any specific element that must remain untouched]

STYLE REFERENCE VOCABULARY

Use these as style anchors. Mix and match per project:

Photography Styles 35mm Kodak Portra 400 · medium format film · Polaroid instant · daguerreotype · Magnum street photography · Annie Leibovitz editorial · fashion editorial Vogue 1994

Rendering Styles Unreal Engine 5 photorealism · Pixar 3D animation · Studio Ghibli watercolor · brutalist graphic design · bauhaus flat vector · risograph print · pencil sketch on textured paper

Lighting Styles (Named) Rembrandt 45° triangle · Chiaroscuro deep shadow · volumetric god rays · butterfly/Paramount beauty · hard split light · rim backlight halo · golden hour backlight · practical neon glow · high-key commercial · low-key noir

Lighting Moods (Descriptive) golden hour warm backlight · blue hour city glow · harsh desert noon · candlelit intimacy · clinical cold fluorescent · neon-soaked rain reflection · volumetric fog rays · overcast diffused flat light · tungsten warmth vs. cool daylight contrast

EXAMPLE BANGER PROMPTS

🏆 Commercial Product

🏆 LinkedIn Headshot

🏆 Live Data Infographic

🏆 Cinematic Scene

🏆 Text-Heavy Social Post

🏆 Architecture Visualization

🏆 Wild Creative

RAPID PROTOTYPING CHECKLIST

Before generating, confirm:

  • [ ] Subject is specific — not generic
  • [ ] Camera and lens named
  • [ ] Lighting direction and quality specified
  • [ ] Campaign context or purpose stated
  • [ ] Any embedded text follows the Text Distance Rule
  • [ ] Search grounding triggered if real-world accuracy needed
  • [ ] Thinking mode enabled if layout is complex
  • [ ] Resolution and aspect ratio stated at the end
  • [ ] For edits: change and preserve clauses both present

Nano Banana 2 is a language model that generates images. Talk to it like a creative director. The more context, intent, and specificity you provide, the closer the output gets to a professional shoot.

Beginning-Willow-801 · 2026-03-03 · 1 point

A humorous but photorealistic macro image of tiny 3-inch-tall people living and adventuring inside a large salad bowl. A group of miniature hikers in proper hiking gear treks across a lettuce leaf landscape, another figure rock-climbs the face of a cherry tomato, a tiny couple picnics on a crouton. Everything in the salad bowl is at normal salad scale; the people are simply very small within it. Real macro food photography lighting: soft top-down diffusion, product-quality crispness. Shot on Sony A7R V, 100mm macro f/5.6. 1:1, 4K.

<image>

Beginning-Willow-801 · 2026-03-03 · 1 point

A photorealistic image of an absolutely enormous rubber duck — 15 stories tall — floating serenely in San Francisco Bay with the Bay Bridge and city skyline behind it. The duck is perfectly yellow and clean, its proportions exactly those of a normal bath toy, simply scaled to building height. It creates a significant wake around it. Morning fog rolls around its base. A few news helicopters orbit it. Shot from a ferry-level angle, 200mm telephoto bringing the bridge and duck together. Search for accurate San Francisco Bay bridge reference. 16:9, 4K.

<image>

Beginning-Willow-801 · 2026-03-03 · 1 point

A luxury apartment interior design concept where the open-plan living room floor is an impossibly large, perfectly formed New York-style pepperoni pizza — the cheese stretching slightly in the foreground as if freshly arrived. The rest of the apartment is perfectly normal and expensive: a white sectional sofa floats on the pizza floor, a marble kitchen counter runs along the back wall, a gallery-quality painting hangs. Pizza steam rises gently. Warm apartment window light. Interior design photography lighting and composition. 16:9, 4K.

<image>

Beginning-Willow-801 · 2026-03-03 · 1 point

A photorealistic studio portrait of a red fawn french bulldog in a sharp navy blue courtroom suit, white dress shirt, and striped tie — dressed as a defense attorney. The dog sits at a law office desk with legal briefs, a gavel, and a framed law degree on the wall behind reading 'Harvard Law School.' Expression: focused and professionally authoritative. Three-point studio lighting: key softbox left, fill right, rim light behind. Shot on Hasselblad X2D, 135mm f/4. The humor is entirely in the absolute seriousness of the composition. 3:4, 4K.

<image>

Beginning-Willow-801 · 2026-03-03 · 1 point

A fun, respectful photorealistic image of Albert Einstein — recognizable trademark white hair and mustache — in 1950s surf shorts, successfully riding a perfect breaking wave in California. His expression shows genuine delight and concentration. The surfboard has 'E=mc²' written in chalk on its deck. A few other surfers are visible in the background. The wave and ocean are completely photorealistic, as is Einstein's physical form. Shot in the style of 1960s surf photography. Natural California beach light at midday. 16:9, 2K.

<image>

william_luckybob · 2026-03-10 · 1 point

I encountered some difficulties when implementing actual product image replacement.
In practical use, the results generated by Google AI Studio were very satisfactory when rendering the product onto the desired background.
However, when I called the API using Nano Banana 2, the generated images were extremely stiff and looked like they were just directly overlaid.
I used exactly the same prompt and parameters.
Could this be because Google AI Studio retains conversational memory?
In Google AI Studio, I first reverse-engineered the prompt from the scene image, then revised and wrote my own prompt, which I passed in together with my product image.
In the API call, I passed both images to the API.

Could you share some tips for that situation?

Beginning-Willow-801 · 2026-03-03 · 0 points

A stunning impossible composite photograph that captures an entire day's passage from sunrise to midnight in a single image. The sky is divided into vertical bands that transition from left to right: dawn pale pink, early morning cool blue, midday bright white-blue, golden hour amber, sunset magenta-orange, deep blue dusk, and full night with stars and moon. The cityscape below is identical throughout, lit consistently from its own lit windows at night. Shot on Canon R5, 24mm f/8. The bands fade seamlessly into each other. 16:9, 4K.

<image>

Beginning-Willow-801 · 2026-03-03 · 0 points

A photorealistic image of a large clear glass bottle — sealed, sitting on a wooden table — that contains a complete miniature city inside. Through the glass, you can see tiny lit skyscrapers with actual windows, a river with bridges, parks with tiny trees, and streets with cars the size of pinheads. The city inside appears to be nighttime with all its lights on; the room outside is daytime. The bottle glass curves and distorts the view slightly, adding to the believability. Shot on Canon R5, 85mm macro f/2.8. 1:1, 4K.

<image>

Beginning-Willow-801 · 2026-03-03 · 0 points

A stunning surreal photograph of a single ornate wooden door — completely free-standing, no walls around it — in the middle of an English countryside meadow. The door is open, and through the doorway is a completely different reality: a desert landscape at golden hour with red sandstone mesas. The meadow with green grass and wildflowers surrounds the door; the desert interior light pours out across the grass, creating a believable interplay of two different lighting environments meeting at the threshold. Shot on Sony A7R V, 35mm f/8. 16:9, 4K.

<image>

jentravelstheworld · 2026-03-04 · 0 points

Wow!