I built Nano Banana 2, an AI image generation platform powered by Google's Gemini 3.1 Flash Image   
  model.                                                                                               
                                                                                                       
  The main problems I wanted to solve:                                                                 
  - Text rendering in AI images has always been broken. This model gets it right ~90% of the time.     
  - Most generators cap at 1-2K. This does true 4K output.                                             
  - No character consistency across images. This tracks up to 5 characters + 14 objects.               
                                                                                                       
  Tech stack: Next.js 16, React 19, TypeScript, Drizzle ORM, PostgreSQL, Cloudflare R2.                
                                                                                                       
  4 models available depending on your needs — from $0.039/image (budget) to $0.134/image (pro).       
  Credits never expire. Free credits on signup.                                                        
                                                                                                       
  Live at https://ainanobanana2.pro                                                                    
                                                                                                       
  Happy to answer any questions about the tech, the Gemini API integration, or the business model

Comments URL: https://news.ycombinator.com/item?id=47196085

Points: 1

# Comments: 0

I built ChibiGenerator, a small web app that generates chibi-style characters from photos or text prompts using AI.

I noticed that many existing avatar or chibi tools either rely on fixed templates or require illustration skills. I wanted something simple: upload a photo, pick a style, and get a clean, usable chibi character in seconds.

The product is currently focused on: - Photo-to-chibi and text-to-chibi generation - Multiple chibi-style outputs - High-resolution images suitable for reuse - A very lightweight, no-friction UI

It’s live and being iterated on based on early user feedback. I’d really appreciate any thoughts on output quality, usability, or potential use cases I may be missing.

Thanks for checking it out.

Comments URL: https://news.ycombinator.com/item?id=46870969

Points: 1

# Comments: 0

It updates every day automatically, includes step-by-step clues, and has a minimal UI with no ads or tracking.

Made it because I couldn’t find a clean, fast reference for Pinpoint players. Feedback on speed, clarity, or features would be appreciated.

Comments URL: https://news.ycombinator.com/item?id=46616681

Points: 1

# Comments: 0

Comments URL: https://news.ycombinator.com/item?id=45865606

Points: 1

# Comments: 0