Nano Banana Pro - Comprehensive Prompt Collection
Open Source License
This project is licensed under the MIT License.
Foreword: Google Releases Nano Banana Pro, Ushering in a New Era of AI Image Generation
OpenAI faces its darkest hour.
Google's AI offensive shows no signs of weakening. While Gemini 3 Pro recently targeted the "frontend" domain, today it's the design industry's turn.
The newly released Nano Banana Pro (Gemini 3 Pro Image) delivers another powerful punch in image generation capabilities. The jobs of junior designers may be in jeopardy.
I. Core Features Overview
- Resolution Support: Outputs up to 4K resolution images
- Multi-round Editing: Supports conversational, multi-round image editing workflows
- Multi-image Composition: Combines up to 14 input images into 1 output image
- Search Enhancement: Integrated Google search capabilities for more precise, up-to-date knowledge support
- Deep Thinking & Logical Reasoning: Performs physics simulation and logical reasoning before image generation, not just simple imitation
- Cross-modal & Multilingual: Supports generating text in multiple languages, one-click translation and localization
- Ultra-long Context: 64K input token limit, understands extremely long text prompts
- Character Consistency: Maintains appearance consistency of up to 5 characters across multiple images
- Professional Creative Control: Adjust camera angles, lighting, styles, color grading through text instructions
- AI Transparency: All generated content includes SynthID digital watermarks for source verification
II. Complete Collection of Official and Creator Test Cases
Part 1: Core Capabilities Demonstration
1. Character Consistency and Mouth Shape Accuracy
Prompt: Please draw a four-panel image showing the same young man wearing a bamboo hat pronouncing "wǒ" (我), "shàng" (上), "zǎo" (早), "bā" (八) respectively. The character's appearance should remain consistent, mouth shapes accurately correspond to each pronunciation, with unified overall style, 16:9, 4K
2. Manga Coloring and Translation
Prompt: Translate the text on the image to Chinese and color it, keep everything else unchanged
3. International Poster Translation and Re-layout
Prompt: Translate all English text on the poster to Chinese
4. Complex UI Interface Replication (Chat History)
Prompt: Generate two vertical WeChat group chat interface screenshots, with group members discussing Nano Banana Pro's release
5. Long Text and Ancient Painting Style Combination
Prompt: Generate a 4K ancient painting with the text: "明月几时有？把酒问青天。不知天上宫阙，今夕是何年。我欲乘风归去，又恐琼楼玉宇，高处不胜寒。起舞弄清影，何似在人间。转朱阁，低绮户，照无眠。不应有恨，何事长向别时圆？人有悲欢离合，月有阴晴圆缺，此事古难全。但愿人长久，千里共婵娟。"
6. IP Character Fusion Creation
Prompt: Doraemon and Li Bai drinking together under the moon. Bright moon hanging high, ancient pavilions, Doraemon wearing Tang dynasty clothing, Li Bai holding a wine pot, stone table with wine vessels, ethereal atmosphere, mixed Chinese-Japanese art style, exquisite details
7. Search Enhancement: Real-time Information Visualization (Travel Planning)
Prompt: Generate a visualization image showing a 2-day travel itinerary in Guangzhou
8. Search Enhancement: Real-time Information Visualization (Weather Info)
Prompt: Search for real-time weather information in Guangzhou and create a Chinese pop art style infographic, 4:3
9. Search Enhancement: Real-time Weather UI Design
Prompt: Help me search for current weather information in Beijing (20251120) and place it in a weather UI design mockup
10. Search Enhancement and Image Editing Combination
Prompt: Search for Beijing's weather today (20251120) and a positive morning news item. Convert my photo to 'American retro poster' style. Add huge artistic text of today's date (Chinese) at the top, and format the news summary and weather tips at the bottom in newspaper layout style
11. Multilingual Journal Creation
Prompt: Help me generate a collage-style travel journal from Guangxi, recording行程 and photos along the way, with text in Chinese, Korean, and English
12. Multi-image Input and Handwritten Diary Style
Prompt: Simulate a handwritten diary about today on slightly textured paper (off-white or light brown). All images arranged in collage style on one page, keeping images consistent with originals including: handwritten font describing today's activities, one or two attractive slogans, introductions to several photos, red or bright pen marks or arrows pointing to recommended places/activities. Simple doodle-style illustrations related to image features, current date and Beijing weather, and a hand-drawn character. Overall feel should be a lively, interesting personal diary carefully created by someone who loves life
13. Chinese Font Design
Prompt: (Example shows comprehensive effects of various font design prompts)
14. Product Poster Design (Single Image)
Prompt: Design a product poster for this beige leather phone case. Super close-up of the case standing sideways, clearly showing the fine texture of vegan leather and the contrasting texture of orange metal camera ring, with magnetic card pocket of same color attached to back. Beige background with brown transparent light gauze surrounding, decorated with several fern plants, agarwood dead wood and two lily of the valley as foreground accents. 4K ultra-high definition image quality. Still life photography, dark atmosphere, light ray tracing, premium soft light. Poster title at top: "AATTT STUDIO", ultra-thin serif font. Small text at bottom: "Vegan Leather | Magnetic | Minimalist". Artistic signature corner tag: "AATTT | Design"
15. Product Poster Design (Multiple Images)
Prompt: Design a product poster for these two aromatherapy products. Super close-up of two aromatherapy products together, texture clear. Beige background with brown transparent light gauze, fern plants, agarwood dead wood, two lily of the valley. 4K ultra-high definition image quality. Still life photography, dark atmosphere, light ray tracing. Poster title at top: "Kunlun Boiling Snow", ultra-fine text. Small text at bottom: "Agarwood | Lily of the Valley | Herbal". Artistic signature corner tag: "To Summer |观夏"
16. Product Scene Rendering
Prompt: Help me place this aromatherapy product in an interior space matching the design style, master photography, close-up shot
17. Multi-object Scene Fusion
Prompt: Place these furniture harmoniously in one room
18. Portrait Fine Editing and Beauty Enhancement
Prompt: Light beauty enhancement, face slimming, hairstyle change to American spikes, hairline forward, glasses changed to sunglasses
19. IP Copyright Content Generation (Cross-cultural Fusion)
Prompt: Jerry Mouse dressed in the iconic yellow tiger skin skirt, big red cape, and golden armor from the animated version of "Havoc in Heaven", wearing a phoenix-wing purple-gold crown, holding a golden cudgel, exaggerated and divine facial expression, background of heavenly palace pavilions or Flower-Fruit Mountain Water Curtain Cave, strictly following the classic style of Shanghai Animation Film Studio's "Havoc in Heaven", rich colors, smooth lines, full of traditional Chinese ink painting and heavy-color meticulous painting charm
20. IP Copyright Content Generation (Relationship Diagram)
Prompt: A journal-style image showing character relationships and power dynamics from the TV series Game of Thrones, with character avatars
21. IP Copyright Content Generation (Game Screenshots)
Prompt: 3D realistic-style Pokémon-themed MMO game screenshot
22. White Model Coloring and Rendering
Prompt: Complete material addition and coloring for this figure, and change surrounding environment to match the character setting
23. Manga Translation and Coloring (Complex Scene)
Prompt: (Example of translating and coloring complex One Piece manga pages)
24. Acid Design/Y2K Style Poster Conversion
Prompt: Please analyze this photo I uploaded and convert it into a modern graphic design poster. Follow these steps: Subject processing: First, precisely cut out the subject from the original background as the core of the poster. Background transformation: Replace the original background with an abstract, frosted glass texture background. This new background should retain the original background's color tendencies (like green and brown) but completely blur specific details. Core composition: Behind the subject, add a striking solid color rectangle block. Extract the color from the brightest color in the photo. Add main title text related to the photo subject, using a bold, designer sans-serif font. Most importantly, place this title layer behind the subject to create spatial perception of the subject obscuring the text. Detail addition: Add some small, informational text related to the image content at different positions to create visual hierarchy. At the bottom of the flower stem, add some transparent tape material to make the flower look like it's pasted on the poster, increasing realism and collage feel. Style unification: Finally, ensure the entire poster follows Y2K aesthetics or acid design, with rich textures and layers. Let the title and color blocks have some transparency fusion with the background's glass texture for overall coordination
25. UI Design Style Migration
Prompt: Imitate this design's style and key element expression, help me generate a weather app UI design mockup
26. Ultra-long Text Ancient Style Scroll
Prompt: At the top of the image, a long scroll with free calligraphy writing the complete poem "Song of the Pipa" by Bai Juyi, starting with "By the Xunyang River at night seeing off guests, maple leaves and reed flowers autumn rustling [full Pipa Song text] who cries most in the audience? Jiangzhou Prefect's green shirt is wet", while below using Chinese ink painting style to draw the corresponding scene for this poem
27. WeChat Group Chat UI Replication (Empresses in the Palace Version)
Prompt: Generate a WeChat mobile chat interface screenshot, strictly following WeChat's interface UI and layout, left dialogue bubbles are gray, right dialogue bubbles are green, with the character Zhen Huan speaking on the right. Group chat name: Forbidden Palace Gossip Group. Chat content:... (see original text for details)
28. Cross-IP Game Scenes
Prompt: Generate an original Genshin Impact real-time battle image, with the Traveler protagonist fighting boss Elon Musk, Musk standing on a hovering rocket armor firing energy cannons, image in 3D high-saturation fantasy Genshin Impact game style
29. Super IP Character Group Photo
Prompt: Sun Wukong from Chinese animation together with Luffy, Naruto Uzumaki, Tanjiro Kamado, Mikoto Misaka, Rem, Kirigaya Kazuto, Asuka Shikinami, Tony Tony Chopper, Madoka Kaname, Anya Forger, Ichigo Kurosaki, Eren Yeager, Ai Haibara and Hatsune Miku and Nezha, Li Bai, Tang Monk, SpongeBob, Crayon Shin-chan, Disney princesses standing together, eclectic group photo, collective grand photo; background is endless grassland, blue sky and white clouds, overall bright and fresh anime style
30. Educational Infographic Creation
Prompt : Create an educational image in Chinese explaining the detailed principles of photosynthesis in plants
Prompt : Design a "Classic Optical Experiment: Newton's Prism Light Splitting" storyboard-style educational poster: left side shows incident white light, middle is glass prism, right side accurately disperses seven-color light bands by wavelength order; small box below recreates "light dispersion" formula n = c/v; overall 4:3 ratio HD vector, dark gray background, sans-serif font
Prompt : Draw a "Human Genome 23 Chromosome Pairs Overview" infographic in Chinese: vertically arrange chromosome bars according to actual length proportions, use different gradient colors to distinguish autosomes and sex chromosomes, list representative genes related to known diseases/traits for each pair (like HBB, BRCA1)
31. Pet Photo Editing and Brand Collaboration
Prompt : Help my cat wear the Zootopia rabbit officer cosplay outfit, including head hat and clothes, reasonably fit the small cat, remove obstacles in front of the cat, maintain original composition and realistic style
Prompt : Create a co-branded poster themed "77 (cat name) X KFC" centered around the same pet (appearance, color patterns etc. absolutely consistent with uploaded image) ... then freely add appropriate Chinese content to this poster
32. Poster Style Migration and Content Replacement
Prompt : Reference the image style, change theme color to pink, replace basket with makeup case, and change all elements and text about fresh fruits to beauty elements and text content
Prompt : Reference the style, color and composition of the image, generate a 2026 Year of the Horse Spring Festival poster, maintaining consistent font
33. Multi-image Input: Product and Poster Style Fusion
Prompt : Reference Image 1's poster style, create a poster for Image 2's laundry detergent product, maintaining Image 2's product consistency, with all elements and text content related to Image 2's laundry detergent
34. Line Art Coloring and Detail Enrichment
Prompt : Based on this line art, generate a detailed urban design rendering
35. Perspective Conversion
Prompt : Change the four-panel scenes to overhead top-down perspective
36. Manga Translation and Coloring (Sequential Operation)
Prompt : Color this manga page and translate to Chinese, place in original positions, maintaining composition and image detail consistency
37. Continue Writing Book Page Content
Prompt : Generate the next page of this book, correct page position, continuous text content, and maintain clear text
38. Math Problem Solving Process Visualization
Prompt : Draw a draft paper with the solution to this problem (input is math problem image)
39. Paper/Long Text to Whiteboard Diagram
Prompt : (Input long text or PDF content) Convert... into a detailed whiteboard photo
40. Product Multi-scene Rendering
Prompt: (Jay Chou CD example, rendered in different scenes)
- 90s Taipei old record store wooden shelf...
- CD floating in mid-air, surrounded by broken glass fragments...
- Putting CD into CD player...
41. Bead Sprite Creation
Prompt : 3D bead sprites, box containing exquisite figures.
Prompt : (Example) Bead sprite puppy
Prompt : (Example) Bead sprite color cards and kitten
Part 2: UI/UX and Interface Replication Special Tests
1. Classic Operating System Interfaces
Windows XP: Accurately replicate Windows XP desktop environment. Requirements: classic blue-green background, 'Start' button in bottom left, open 'My Computer' window, file list fonts and icons must be faithful to XP style
Windows 11: Generate Windows 11 desktop screenshot. Requirements: three windows open simultaneously: semi-transparent 'Settings' window, browser window, and File Explorer window. Correct window shadows and hierarchical relationships, accurate Chinese text in window title bars
Mac OS 9: Replicate Mac OS 9 desktop screenshot. Requirements: display black and white or low-color pixelated icons, system menu bar, and two overlapping gray windows. Fonts and jagged window edges (aliasing) matching early display characteristics
macOS (Latest): Replicate latest macOS desktop. Requirements: top menu bar clearly showing WiFi icon, time (9:41 AM), and a popping system notification (content: 'Software update completed', text must be accurate)
2. Data and Application Interfaces
SaaS Dashboard: Generate SaaS platform data dashboard. Requirements: include high-precision bar chart, pie chart, and at least three different colored progress bars. Charts and progress bars must have accurate percentage numbers (e.g.: 78%, 45%)
Data Dashboard: Help me create a new energy virtual power plant data dashboard for leadership presentation
APP Mockup: I want to develop an APP similar to "Xiaoyuzhou", please help me draw a mockup, all requirements meeting APP mockup standards
PPT Slide: I want to make a PPT slide about AI Agent application scenarios, please help me design, tech theme required, text in Chinese
3. Game Interfaces (HUD)
Sci-fi RPG: Generate sci-fi RPG game battle interface. Requirements: top-left corner shows player status bars (Health: 25/100, Mana: 150/200...); bottom-right corner shows six skill icons, three in Chinese cooldown countdown state (e.g.: 'Cooldown: 3.5s')
RTS: Generate RTS game battle interface. Requirements: top-right corner clearly displays three resource quantities (e.g.: 'Ore: 1258'...); bottom-left shows six unit icon cards...
FPS: Generate FPS game battle screen. Requirements: screen center has HUD scope; bottom-right shows weapon ammo count (e.g.: '28/120'); screen must have Chinese 'Headshot' prompt effect...
City Sandbox: Generate city sandbox game isometric top-down screenshot. Requirements: mouse hovering above Chinese-named commercial building (e.g.: 'Li's Mall'), building popup shows Chinese info bubble displaying 'Current Income: +¥12,500'
MOBA: Generate MOBA game bottom center HUD. Requirements: five skill icons clearly visible, two in Chinese cooldown countdown state; also shows six equipment slots...
Board Game 'Catan': Generate high-precision replica of 'Catan' board game
'JX3 Online': Generate 'JX3 Online' battle interface screenshot. Requirements: ...bottom shows skill shortcut bar, three skills in Chinese cooldown state (e.g.: 'Cooldown: 12s')...
'League of Legends': Generate 'League of Legends' battle interface. Requirements: ...bottom center shows five skill icons, ultimate (R key) in cooldown countdown state (e.g.: '58s')
Night Vision FPS: Generate night battle screenshot. Requirements: night vision mode...screen center Chinese warning text shows: 'Warning: Battery Low!'
Sniper Scope: Generate close-up of sniper aiming at target through high-magnification scope. Requirements...screen top-right shows Chinese wind speed/distance data (e.g.: 'Distance: 350m, Wind Speed: 5m/s')
Wuxia Game Cutscene: Generate wuxia game story cutscene screenshot...bottom shows Chinese widescreen subtitles: 'Ten years of grievances, settled today!'
Part 3: Content Creation and Stylized Applications
1. Social Media Styles (Xiaohongshu Covers)
Food Blogger: Generate high-saturation food note cover. Theme: '😭 Stayed up all night organizing! 3-day lazy weight loss meal plan'...Chinese emoji (😭) must be clear and not garbled
Beauty Blogger: Generate beauty tutorial cover. Requirements: portrait close-up using 'Korean clean and cold' filter...bottom add Chinese title: 'Beginner must-see! One-step yellow tone removal for natural makeup'...
Tech Blogger: Generate digital product review note inner page. Requirements...side has Chinese parameter comparison table...table must have accurate Chinese characters and numbers/units
AI Blogger: Generate AI drawing tutorial note cover. Theme: '😭 Stayed up all night organizing! 30 Midjourney core keywords, beginners can create amazing images!'...
2. Visual Novels and Interactive Scenes
Modern Romance: Generate modern campus-style visual novel scene...dialogue box shows male lead's Chinese dialogue: 'Why don't you believe me?', male lead expression should be 'anxious', female lead expression 'doubtful'
Ancient Wuxia: Generate ancient wuxia scene...dialogue box shows third character's Chinese line: 'This object is extraordinary, who will safeguard it?'...
Fantasy Adventure: Generate fantasy adventure style visual novel interface...bottom clearly shows two Chinese plot option buttons: '1. Throw dagger' and '2. Summon contracted demon beast'
Sci-fi Mech: Generate sci-fi mech themed interactive scene. Requirements: top-right corner has Chinese status HUD showing: 'Pilot: Luo Bing', plus a 'Mental Power' progress bar...
3. Complex Text and Ancient Book Replication
'Dream of the Red Chamber': Generate an open book, left page is the opening paragraph of Chapter 1 of 'Dream of the Red Chamber' (about 100 characters), right page is vertical traditional Chinese calligraphy style summary...
'The Analects': Generate ancient thread-bound book double-page text close-up. Requirements: text accurately replicates 'Learning Chapter' full text of 'The Analects'...
'Lantingji Xu' Stele: A stele carved with sutras, inscribed with complete 'Lantingji Xu'. Text must run left to right along the stele, maintaining running script brush intent
Chinese Capital Number Calculator: An electronic calculator screen showing calculation in Chinese characters: 'Three thousand one hundred forty-one multiplied by two thousand seven hundred eighteen equals (accurate result in Chinese capital numbers)'
Ancient Painting Inscription: Generate Southern Song Dynasty blue-green landscape painting scroll. On the right side of the scroll, a running-regular script Chinese inscription with Su Shi's 'Inscription on West Forest Wall' complete poem
4. Manga Creation and Style Conversion
Complete Manga Processing: Turn this image into color, translate to Chinese and place in corresponding speech bubbles, keep everything else unchanged, finally give me as image. (Operating on original Demon Slayer image)
(Original manga)→(Translation and coloring)
(Style conversion)→3D plush/mosaic/relief
5. Poster Design
Movie Poster 'Dead Poets Society': 'Dead Poets Society' movie art poster...main title in flowing artistic handwritten font 'Dead Poets Society'...
Movie Poster 'Legend of the Monkey King': Animated movie 'Legend of the Monkey King' art poster, Chinese landscape painting style...theme in handwritten artistic font 'Legend of the Monkey King'...
Event Poster: I'm holding an event themed 'AI Empowers Invention and Innovation'...please help me design a promotional poster...
6. Knowledge Graphs and Infographics
Yingxian Wooden Pagoda Structure: Generate structural explanation diagram for Yingxian wooden pagoda in Datong, Shanxi, with detailed Chinese text, clear Chinese characters
Su Embroidery Craft: Generate detailed explanation diagram for Su embroidery craft, with detailed Chinese knowledge analysis
'One Hundred Years of Solitude' Character Relationships: Draw character relationship diagram from 'One Hundred Years of Solitude', relationships shown in Chinese
Lithium Battery Principle: Lithium battery working principle diagram, labeled in both Chinese and English
7. Personal and Educational Applications
Personal Business Card: ...please generate a personal introduction image for me, I need to present externally...
Kung Fu Illustration: Recently learning Eight Brocade Qigong, have Trump teach me with Chinese diagram explanations
Ancient Poetry Illustration: Please illustrate the poem "Plucking chrysanthemums by the eastern fence, I calmly see the southern mountain"
IV. Advanced Features and Applications
Converting Articles to Styled Infographics
Due to enhanced long text rendering capabilities, current popular tests focus on generating various infographics - from dense knowledge explanations to classic ancient poetry filling blackboards and walls, or math problem solutions.
For example, converting input text verbatim into glossy magazine content:
Image credit: X@fofrAI | Prompt: Put this whole text, verbatim, into a photo of a glossy magazine article on a desk, with photos, beautiful typography design, pull quotes and brave formatting. The text: [...the unformatted article]
Some users directly provide Markdown documents or PDFs, asking Nano Banana Pro to convert them into infographics, resulting in design-friendly, information-accurate images.
Image credit: X@tobi | Prompt: Make this markdown transcript into a infographic
Because Nano Banana Pro can now use Google search, and reasoning and understanding capabilities have greatly improved, when generating infographics, if requirements aren't too strict, simply telling it what kind of infographic to create can achieve results without worrying too much about prompt structure.
Prompt: Generate a 2026 China holiday schedule infographic, all in Chinese, 4K quality, 16:9
For example, when asking it to generate next year's holiday information, I saw Gemini's reply included clear use of Google search results, with multiple official announcement webpage links.
Many beautiful infographic tests have relatively simple prompts, basically "help me generate an xx infographic".
A fun busy town infographic, showing animals working in a busy town and introducing their different jobs.
Image credit: X@unsoldwill | Prompt: Make a fun busy town infographic of animals going to work in a busy town showcasing different jobs.
A Stephen Biesty style double layer neural network diagram.
Image credit: X@jon_barron | Prompt: Generate a diagram of a two-layer neural network in the style of Stephen Biesty
This user also gave this image to Google's Veo 3.1 to animate it.
Convert content-dense documents into clear, scannable whiteboard-style infographics.
Image credit: X@denilgabani | Convert the attached research paper into a single whiteboard-style image. Break down all key concepts, diagrams, insights, and relationships in a way that is easy to understand at a glance. Add clear visual notes, arrows, highlights, summaries, and clever student-style annotations so I can quickly grasp the full paper and take notes from it.
Generating Exciting Comic Strips
Combining long text rendering and strong consistency maintenance, besides rendering text-heavy infographics, Nano Banana Pro is perfect for creating comic strips. It now supports generating multiple photos in a single request.
Some users just uploaded one image, then input "create and draw a multi-panel comic based on this image" prompt, and got multiple self-written plot comics, and they continue updating subsequent plots.
Swipe left for more content, Image credit: X@DeryaTR_ | Prompt: Create and draw a multi-panel comic based on this image.
There are also prompts to generate several pages of comics starring someone's life. Nano Banana Pro can enable web search, retrieve this person's information, find their X account, and generate comics based on their X account's posted content.
Swipe left for more content, Image credit: X@skirano (Pietro Schirano) | Prompt: Generate a few pages of comic starring the life of Pietro Schirano
If you also have public social media platforms, try letting Nano Banana Pro generate comics from your past week's life.
We tried having Nano Banana Pro generate life comics for Musk and Altman. In the Gemini web version, it gave us three images at once, and you can see from the comic content that it really does have world knowledge.
Swipe left for more content, Prompt: Generate several real-life comic images starring Elon Musk and Sam Altman's lives, all in Chinese
As for the previous generation Nano Banana's consistency capabilities, the Pro version still performs excellently. Like dressing a model in different clothes, maintaining model and clothing consistency; also like putting some logos on different peripherals, or making small element replacements to existing product images, Nano Banana Pro performs very well.
In Google AI Studio's official case studies, multiple infographic generation and comic production examples are also given. They built multiple projects using the Nano Banana Pro model.
Experience address: https://aistudio.google.com/apps?source=showcase\&showcaseTag=nano-banana
One of them only requires us to upload an article link, and it can generate four different style infographics, including fun type, simple type, dark tech style, and modern editorial style. Also just giving it a vague concept or theme can generate a style-varied infographic for us.
Swipe left for more content, Project introduction Demo
4K wallpaper generation, arcade game real-person experience, and more brand-appropriate logo sticker tools are all available as ready-made projects on Google AI Studio.
Golden Gate Bridge in snow 4K wallpaper
We used the "Infinite Heroes" small project to generate a comic strip about Musk and Altman, and while previewing the comic strip, we could also choose the plot direction to determine the next comic content.
Swipe left for more content, downloaded comics are PDF files
However, the Nano Banana Pro model API is not free to call, so these six AI Studio projects all require binding payment methods to use.
But these projects all provide source code, and we can modify them based on their foundation. By checking the underlying code, we can also see the underlying prompts.
We extracted one project's prompt, which is to generate four styles of infographics from input links.
First, we can use the following prompt to summarize web link content, replacing u r l ∗ ∗ w i t h t h e a c t u a l a r t i c l e l i n k t o a n a l y z e . ∗ ∗ {url}** with the actual article link to analyze. ** url∗∗withtheactualarticlelinktoanalyze.∗∗{language} can be directly replaced with Chinese to present in Chinese.
You are an expert Information Designer. Your goal is to extract the essential structure from a web page to create a clear, educational infographic.
Analyze the content at this URL: ${url}
TARGET LANGUAGE: ${language}.
Provide a structured breakdown specifically designed for visual representation in ${language}:
- INFOGRAPHIC HEADLINE: The core topic in 5 words or less (in ${language}).
- KEY TAKEAWAYS: The 3 to 5 most important distinct points, steps, or facts (in ${language}). THESE WILL BE THE MAIN SECTIONS OF THE IMAGE.
- SUPPORTING DATA: Any specific numbers, percentages, or very short quotes that add credibility.
- VISUAL METAPHOR IDEA: Suggest ONE simple visual concept that best fits this content (e.g., "a roadmap with milestones", "a funnel", "three contrasting pillars", "a circular flowchart").
Keep the output concise and focused purely on what should be ON the infographic. Ensure all content is in ${language}.
After getting this article summary, we can start image generation.
Create a professional, high-quality educational infographic based strictly on this structured content plan:
${structuralSummary}
VISUAL DESIGN RULES:
-
${styleGuidelines}
-
LANGUAGE: The text within the infographic MUST be written in ${language}.
-
LAYOUT: MUST follow the "VISUAL METAPHOR IDEA" from the plan above if one was provided.
-
TYPOGRAPHY: Clean, highly readable sans-serif fonts. The "INFOGRAPHIC HEADLINE" must be prominent at the top.
-
CONTENT: Use the actual text from "KEY TAKEAWAYS" in the image. Do not use placeholder text like Lorem Ipsum.
-
GOAL: The image must be informative and readable as a standalone graphic.
The image generation prompt is relatively simple. We replace the previous article link analysis result into the s t r u c t u r a l S u m m a r y ∗ ∗ p a r t ; t h e n f o r ∗ ∗ {structuralSummary}** part; then for ** structuralSummary∗∗part;thenfor∗∗{styleGuidelines}, we can choose any of the four styles mentioned in the official project.
Interesting and fun style Fun & Playful: styleGuidelines = STYLE: Fun, playful, vibrant 2D vector illustrations. Use bright colors, rounded shapes, and a friendly tone.
Simple minimalist style Clean Minimalist: styleGuidelines = STYLE: Ultra-minimalist. Lots of whitespace, thin lines, limited color palette (1-2 accent colors max). Very sophisticated and airy.
Dark mode tech style Dark Mode Tech: styleGuidelines = STYLE: Dark mode technical aesthetic. Dark slate/black background with bright, glowing accent colors (cyan, lime green) for data points.
Modern editorial style Modern Editorial: styleGuidelines = STYLE: Modern, flat vector illustration style. Clean, professional, and editorial (like a high-end tech magazine). Cohesive, mature color palette.
Following this method, we gave Gemini the article about Nano Banana Pro's release and got the following infographics.
Swipe left for more content
Official Intervention: 7 Image Generation Tips
Besides using these already-written prompts, Google has painstakingly provided detailed Nano Banana Pro usage techniques, partly because they're afraid we won't operate properly, leading to misunderstandings about the model. On the other hand, Google really hopes Nano Banana Pro won't just be a fun image generation model - they truly expect it to improve our productivity.
The importance of prompts for getting precise content control, besides the model's own capabilities, the remaining determining factor is the prompts themselves. Google mentions again that a good prompt should clearly include subject, composition, action, location, style, and editing instructions.
Briefly describe the image subject with concise language, for example, specifically explaining what people or things are in the picture? For composition, tell the model how the photo is framed? Is it close-up, wide-angle lens, low-angle shot, or portrait lens, etc.
Example: A fluffy calico cat wearing a small wizard hat, casting magic on a sunlit grassland at sunset, low-angle shot, photorealism
Next is describing the action, what's happening to the subject in the picture? And where it's happening, describing the specific scene. Finally, determine the photo style we want, for example, 3D animation, film noir, watercolor, photorealism, 1990s product photography, etc., different overall aesthetic style choices.
If pursuing more controllable image generation effects, Google also mentioned further improving camera, lighting, and format details to refine prompt content.
For example, current Nano Banana Pro supports custom composition and aspect ratios. We can choose 9:16 vertical posters, 21:9 cinematic wide screens, and other screen formats.
Additionally, in photography and lighting details, Google's reference prompt is to guide shooting like a cinematographer, with terminology as professional as possible, for example, "low-angle shot, shallow depth of field (f/1.8), golden hour backlight creating long shadows, cinematic color grading with soft blue-green tones"; and other prompt language.
Example: A fluffy calico cat wearing a small wizard hat, casting magic on a sunlit grassland at sunset, low-angle shot, shallow depth of field (f/1.8), golden hour backlight creating long shadows, cinematic color grading with soft blue-green tones. Photorealism. 21:9 cinematic widescreen. Title "Cat Magic Casting" displayed at top in cute rainbow font.
When handling text processing tasks, need to clearly specify text content and display position, and corresponding fonts. For chart infographics, emphasize accuracy requirements in prompts and ensure input is correct.
When involving multi-image reference and image editing, need to clearly tell the model each image's role and specific editing needs.
Mastering prompt formulas is just the foundation. Next are some officially recommended specific application scenarios for Nano Banana Pro.
7 Key Application Scenarios
1. AI Text Generation: Previously, AI image generation feared text, producing garbled characters. The new model has made huge improvements in text rendering. We can directly ask it to generate poster titles, complex charts, or detailed model specifications.
Prompt: 制作一张信息图，用中文展示如何制作老北京豆汁儿 (Make an infographic in Chinese showing how to make traditional Beijing fermented soybean milk)
2. Physics and History Knowledge: Based on Gemini 3's reasoning capabilities, this model has richer world knowledge. It can generate logically sound charts or historically accurate scenes.
3. Up to 14 Images Supported! Super Fusion Method: This is probably one of the most powerful features of the new model. It allows us to input 6 to 14 (specific quantity depends on platform) completely unrelated images for AI to fuse and create entirely new scenes.
Using the multi-image input feature, we can have the same character maintain facial and feature consistency across different scenes, and even maintain similarity in group shots. This is perfect for creating comics or storyboards.
Prompt: 将这些图像组合成一张 16:9 格式的适当排列的电影图像，并将模特身上的裙子换成图像中的裙子 (Combine these images into a properly arranged 16:9 format movie image, and replace the model's skirt with the skirt in the image)
4. Precise Size Adjustment: Try different aspect ratios and generate clear visual effects with 1K, 2K, or 4K resolutions on various products.
5. "Point-and-Shoot" Precise Image Editing: Google calls this studio-level control editing for achieving more professional effects.
6. Brand Design Magic Tool: Automatic Logo Application: For designers, compared to first-gen nano banana, its consistency maintenance ability is stronger. Brand companies can have AI print corresponding logos, patterns or artwork on 3D objects, while the printed logos automatically adjust to the original image's natural lighting and texture.
7. Cross-language Seamless Translation: It's not just an image generation tool, it can directly "translate" text within images.
Prompt: 将三个黄蓝罐头上的所有英文文字翻译成韩语，其他内容保持不变 (Translate all English text on three yellow-blue cans to Korean, keep everything else unchanged)
Although Nano Banana Pro is very powerful, in the official documentation, they also honestly list some current limitations. Everyone should pay attention when using them. Mainly some very small text rendering, complex spellings may still have errors.
Additionally, if you use it to generate data charts or infographics, they also remind that manual data verification is necessary. Although AI has logic, it can sometimes confidently talk nonsense - after all, we've all seen how serious AI hallucinations can be.
Finally, when performing extremely complex lighting changes or image fusion, generated images may occasionally have unnatural artifacts.
Currently this model is available on Gemini App and Gemini web version, while on AI Studio and Vertex platforms it requires paid API to experience.
VI. Access Channels
International Access:
Domestic Access Channels:
https://maynorai.top/list/#/home
https://draw.claude-opus.top/list/#/home
https://pro.nanobanana-free.top/
Proxy API:
https://apipro.maynor1024.live/
VII. Related Projects
- awesome-nano-banana : The most comprehensive AI text-to-image prompt website Visit Online - GitHub Repository
This collection is continuously updated with the latest community discoveries and official examples. Contributions are welcome!
For example, converting input text verbatim into glossy magazine content:
Image credit: X@fofrAI | Prompt: Put this whole text, verbatim, into a photo of a glossy magazine article on a desk, with photos, beautiful typography design, pull quotes and brave formatting.
Some users directly provide Markdown documents or PDFs, asking Nano Banana Pro to convert them into infographics, resulting in design-friendly, information-accurate images.
Image credit: X@tobi | Prompt: Make this markdown transcript into a infographic
Comic Strip Generation
Combining long text rendering and strong consistency maintenance, Nano Banana Pro is perfect for creating comic strips. It now supports generating multiple photos in a single request.
Users simply upload an image and input prompts like "Create and draw a multi-panel comic based on this image" to get multi-panel comics with self-written plots.
Google AI Studio Projects
Google AI Studio provides multiple official projects showcasing Nano Banana Pro capabilities, including:
- Link-to-infographic generation (4 styles: Fun & Playful, Clean Minimalist, Dark Mode Tech, Modern Editorial)
- Infinite Hero interactive comic creation
- 4K wallpaper generation
- Arcade game真人体验
- Brand logo tools
Project demo: https://aistudio.google.com/apps?source=showcase\&showcaseTag=nano-banana
IV. Official Prompt Engineering Guidelines
Google emphasizes the importance of good prompts that clearly include subject, composition, action, location, style, and editing instructions.
Basic prompt structure: Briefly describe the image subject, composition (close-up, wide-angle, low-angle shot, portrait shot, etc.), action, scene location, and desired style (3D animation, film noir, watercolor, photorealism, 1990s product photography, etc.).
Advanced controls: For more precise generation, enhance camera, lighting, and format details:
- Custom aspect ratios (9:16 portrait posters, 21:9 cinematic widescreen)
- Professional photography terminology: "Low-angle shot, shallow depth of field (f/1.8), golden hour backlight creating long shadows, cinematic color grading with soft blue-green tones"
- Text processing: Specify text content, display position, and fonts
- Charts and infographics: Emphasize accuracy requirements
- Multi-image references: Clearly specify each image's role and editing requirements
7 Key Application Scenarios
- AI Text Generation: AI can now "write" text instead of creating garbled characters. Generate poster titles, complex charts, or detailed model specifications
- Physics and History Knowledge: Based on Gemini 3's reasoning capabilities, generates logically sound charts or historically accurate scenes
- 14-Image Fusion: Supports inputting 6-14 completely unrelated images for AI to blend into new creations. Excellent for comics and storyboards
- Precise Sizing: Generate 1K, 2K, or 4K resolution visuals with different aspect ratios
- Studio-level Editing: Professional control editing with "point-and-shoot" precision
- Brand Design: Automatic Logo Application: Stronger consistency maintenance than first-gen nano banana. Automatically adjusts logos to natural lighting and texture on 3D objects
- Cross-language Translation: Directly "translate" text within images
Prompt: Translate all English text on three yellow-blue cans to Korean, keep everything else unchanged
V. Limitations and Availability
Current Limitations:
- Very small text rendering, complex spellings may still have errors
- Data charts and infographics require manual verification (AI hallucinations)
- Extremely complex lighting changes or image compositing may occasionally produce unnatural artifacts
Availability:
- Free: Gemini App and Gemini web version
- Paid API: AI Studio and Vertex platforms
VI. Access Channels
International Access:
Regional Access Channels:
- https://maynorai.top/list/#/home
- https://draw.claude-opus.top/list/#/home
- https://nanobanana-free.top/
Proxy API:
📝 Submission Guidelines
We welcome contributions of more high-quality prompt examples to this project!
Submission Requirements
- Complete Content: Include prompts and result images
- Clear Categories: Label case categories (e.g., Character Consistency, Translation & Coloring, Poster Design, etc.)
- Quality First: Prioritize outstanding and representative examples
- Standard Format: Submit according to existing format
Submission Methods
- GitHub Submission: Submit Pull Request directly to this project
- Issue Submission: Submit new cases in GitHub Issues
- Email Submission: Send email to [project@example.com]
Submission Template
markdown
**Case Title**: Brief description of the case content
**Category**: [Character Consistency/Translation & Coloring/Poster Design/Game Design/UI Design/Product Rendering/IP Creation/Traditional Art/Infographics/Style Transfer/Scene Editing/Visual Design]
**Prompt**: Detailed prompt content
**Result Images**: Result image links or attachments
**Notes**: (Optional) Special notes or technique sharing for the case🤝 Contributors
Thanks to all friends who have contributed cases and prompts to this project!
- Welcome to submit Pull Requests to contribute more high-quality cases
- Submit Issues to report problems or suggestions
- Share this project to let more people know about the powerful capabilities of Nano Banana Pro
This collection is continuously updated with the latest community discoveries and official examples. Contributions are welcome!