How are viral trends different from educational and commercial uses of LLM image generation and editing? A privacy-preserving descriptive analysis
Authors: Will Tucker-Ray, Marika Swanberg, Alexander Knop, Marvin Chow, Eric Culbertson, Julian Jacobs, Pritish Kamath, Ravi Kumar, Jason Mothersill, Mikhail Pravilov, Eugene Quan, Yurii Sushko
Organizations: Google Behavioral Economics · Google Research · Google Marketing · Google DeepMind · Google Learning · Google Privacy, Security, & Safety · Google Gemini Data Science
Gemini's Nano Banana models, first introduced in late 2025, led to a shift in the capabilities and use of image generation and editing. We utilize Urania, a Google-developed framework for generating insights about LLM chatbot interactions with rigorous differential privacy (DP) guarantees, to analyze use cases and intents, techniques and ways of prompting, without ever looking at a single image or individual prompt. We select a subsample of 2.1M conversations in early 2026 which utilized Nano Banana models, into roughly 800 clusters in five categories: (1) Social & Viral Sharing, (2) Brand & Marketing Collateral, (3) Educational & Technical Information Media, (4) Product Visualizations & E-commerce, and (5) Architectural & Spatial Designs. We then compare these clusters on 21 features of use, measuring aspects of prompts' (a) editing techniques, (b) artistic style requests, (c) composition, and (d) chatbot interactions. This allows us to describe a baseline of how people are using image generation and editing tools for different use cases---(including growing and emergent use cases which were not possible before Nano Banana)---in a privacy-preserving manner.
Figures & tables
Figure 1: Visualization (generated by Nano Banana) of Urania framework for summarizing chatbot conversations.
Figure 2: The framework’s data flow. Yellow nodes denote non-DP data, green nodes represent operations that are either DP or per conversation, light blue nodes denote private data, and dark blue nodes represent non-private operations.
Figure 3: Conversation volume distribution by country.
Figure 4: Conversation volume distribution by language.
Figure 5: Conversation linguistic properties distribution by category.
Example Social & Viral Image Generation Clusters
• 19th-Century Botanical Atlas Illustrations with Personal Annotations • Crying with Celebrities at Night in Convenience Stores • Generating images of people in Brazil jerseys • Generating Met Gala Outfits and Red Carpet Looks • Generating Football Broadcast Screenshots with Specific Fans • Transforming Photos into Distinct Cartoon Styles • Polaroid-Style Photos with White Curtain Background • Cinematic Modification: Adding Demonic Smoke Entities • Playful Baldness Transformations in Image Editing • Visualizing Age Progression and Regression on People • Personalized Sports Jersey Edits for Photos • Whimsical Cat Image Generation and Editing • Personalized Birthday and Greeting Cards with Text • Transforming Photos into Artistic Oil Paintings
Table 1: Example Social & Viral Image generation cluster titles.
Rank
Artist
1
Rembrandt
2
Salvador Dali
3
Francisco Goya
4
Vincent Van Gogh
5
Leonardo da Vinci
6
Caravaggio
Table 2: Most frequently referenced Artist keywords in Nano Banana prompts.
Figure 6: A co-author in Rembrandt-style portrait (included with permission).
Rank
Sport
1
Motorsports
2
Football (Ambiguous)
3
Baseball
4
Soccer
5
Cricket
6
Swimming
Table 3: Most frequently referenced sports keywords in Nano Banana prompts.
Figure 7: Illustrative social media image based on a sports tweet .
Figure 8: Prevalence of Feature Use in Social and Viral Conversation Clusters.
Product Visualizations & E-Commerce
Brand & Marketing Collateral
• Cake Photography: Focus on Realistic Digital Staging • Generate Realistic Bag Visuals for E-Commerce • Product Visualization & Modification of Bottles for E-Commerce • Packaging Design for Product Visualization & E-commerce • 3D Photorealistic Product Renders with Technical Precision • Product Image Enhancement for E-commerce Display • Product Image Editing: Focus on White Backgrounds • Apparel Visualization on Models and Mannequins • Luxury Jewelry Visualization and Enhancement on Models • Footwear Product Visualization, Modification, and Display • Changing Apparel Colors: Black, White, and More
• Food & Beverage Logo and Brand Identity • Team Logos & Branding with Text Modifications • Refining Logos and Visual Identity with Text and Color Edits • Gaming and Esports Logo Design with Characters • Brochure and Pamphlet Design and Creation with Images • Custom Sticker Design and Print Preparation with Text • Custom Business Card Design and Content Edits • Event-Focused Flyer Design and Text Refinement • Restaurant Menu Design with Food Images and Pricing • YouTube Channel Branding: Logos and Banners • Creating World Cup Themed Sports Graphics and Logos
Architectural & Spatial Design
Educational & Technical Information
• Refining Room and Floor Plan Layouts • Precise Technical Drafting for Complex 3D Models • Detailed Generation of Architectural and Technical Spatial Plans • Detailed Floor Plans: Generation, Modification, and Arrangement • Rendering Interior Staircases with Architectural Design Modifications • Visualizing Plant Arrangements and Garden Layouts • Exterior House Color and Modification Iterations • Architectural Rendering of Doors, Gates, and Windows • Architectural Exterior Modification: Structure, Facade, and Elements • Interior Design: Furniture Placement and Room Redesign • Optimizing Interior Layouts and Storage Solutions
• Visualizing and Explaining Historical and Astrological Concepts • Detailed Medical and Scientific Illustration Requests • Visualizing Complex Organic Chemistry Structures and Reactions • Visualizing Mathematical Concepts and Problem-Solving • Physics Diagrams and Scientific Problem Visualization • Generating Educational Materials and Worksheets on Demand • Educational Poster Creation for Class Projects • Step-by-Step Image Editing and AI Generation • Video Editing and AI Creation Guides • Fitness and Diet: Personalized Exercise and Posture Correction • Hands-on Guides for Repairing and Maintaining Physical Items
Figure 9: Prevalence of features in Architectural & Spatial Design conversations.
Figure 10: Prevalence of features in Brand and Marketing Collateral conversations.
Figure 11: Prevalence of Existing or Original Brands referenced in Brand and Marketing Collateral clusters.
Figure 12: Prevalence of features in Educational and Technical Information Media conversations.
Figure 13: Prevalence of Educator Roles referenced in Educational & Technical Information Media clusters.
Figure 14: Prevalence of Subjects / Domains referenced in Educational & Technical Information Media clusters.
Figure 15: Prevalence of Grade / educational level referenced in Educational & Technical Information Media clusters.
Figure 16: Prevalence of features in Product Visualization & E-Commerce conversations.
Figure 17: Prevalence of text and typography by high level conversation category.
Figure 18: Prevalence of work role ratings by high level conversation category.
Figure 19: Prevalence of workflow phase ratings by high level conversation category.
Figure 20: Summary heat map of binary feature ratings by high level conversation category.
Rank
Country
Prevalence (% Yes)
1
Malaysia
32.2%
2
Singapore
30.1%
3
Estonia
29.1%
4
Finland
29.0%
5
Indonesia
28.7%
6
Taiwan
28.6%
Table 5: Top 10 countries by prevalence of precise positioning prompts.
Rank
Country
Prevalence (% Yes)
1
Somalia
28.4%
2
Malaysia
28.3%
3
India
28.1%
4
Nigeria
27.7%
5
Sri Lanka
27.5%
6
Nepal
24.2%
Table 6: Top 10 countries by prevalence of modifying lighting.
Rank
Country
Prevalence (% Yes)
1
Chile
5.2%
2
Argentina
5.0%
3
United States
4.9%
4
Australia
4.6%
5
Canada
4.4%
6
Mexico
4.3%
Table 7: Top 10 countries by prevalence of imitating art, persons, or characters.
Rank
Country
Prevalence (% Yes)
1
Pakistan
0.7%
2
United States
0.7%
3
Philippines
0.6%
4
Thailand
0.6%
5
India
0.6%
6
Viet Nam
0.5%
Table 8: Top 10 countries by prevalence of meme style or format prompts.
Rank
Country
Prevalence (% Yes)
1
Argentina
2.6%
2
Japan
2.5%
3
Poland
2.5%
4
Chile
2.1%
5
United States
2.1%
6
Canada
2.0%
Table 9: Top 10 countries by prevalence of absurdity or humor.
Rank
Country
Prevalence (% Yes)
1
India
65.5%
2
Iraq
65.3%
3
Somalia
65.0%
4
Nigeria
63.5%
5
Indonesia
63.0%
6
Malaysia
62.6%
Table 10: Top 10 countries by prevalence of people in prompts.
Rank
Country
Prevalence (% Yes)
1
Bangladesh
35.2%
2
Morocco
29.3%
3
Egypt
28.4%
4
Sri Lanka
27.9%
5
Iraq
27.7%
6
Algeria
26.4%
Table 11: Top 10 countries by prevalence of clothing or apparel modification.
Rank
Country
Prevalence (% Yes)
1
Cambodia
21.2%
2
Hong Kong
18.9%
3
Singapore
18.0%
4
Philippines
17.9%
5
Senegal
17.8%
6
Malaysia
17.4%
Table 12: Top 10 countries by prevalence of full reprompts in conversations.
Appendix figures & tables3 assets
Supplementary material from the paper’s appendix.
Appendix
Figure A1: Correlation of feature prevalence across subcategories.
Figure A2: Illustrative “World Cup” social media prompt, generated output, and most frequently referenced countries (image of co-author included with permission).
Figure A3: Illustrative monuments / wonders of the world prompt, generated outputs, and most frequently referenced monuments (images of coauthors included with permissions).
Tsinghua University, Key Laboratory of Pervasive Computing, Ministry of Education, Beijing, China · Microsoft Research Asia, Beijing, China · Department of Computer Science and Technology, BNRist, Tsinghua University, Beijing, China