Summary

Nano Banana Editor is an advanced AI image generation and editing application. It targets creators and developers, offering an intuitive UI for AI image generation and region-aware, conversational revisions powered by Google's Gemini 2.5 Flash Image model. The platform provides a production-ready environment with features like reference images, mask-based editing, and version history for efficient image manipulation.

How It Works

This React + TypeScript application leverages Google's Gemini 2.5 Flash Image model for text-to-image generation and conversational editing. It employs a non-destructive workflow with an interactive canvas (Konva.js) for precise mask painting and region-aware selection. Users guide generation via prompts and reference images, then refine edits using natural language, with a comprehensive undo/redo history tracking all changes.

Quick Start & Requirements

• Install: Clone the repository, cd nano-banana-image-editor, and run npm install.
• Prerequisites: Node.js 18+ and a Google AI Studio API key (set as VITE_GEMINI_API_KEY in .env).
• Run: npm run dev.
• Demo: A live demo is mentioned, but no URL is provided.
• Documentation: Links to Google AI Studio for API key acquisition and official Gemini API documentation are referenced.

Highlighted Details

• AI Creation: Text-to-image generation, real-time prompt tips, up to two reference images, and advanced control over creativity/seeds.
• Intelligent Editing: Conversational modification, region-aware mask selection, style reference images, and non-destructive workflow.
• Project Management: Generation history, variant comparison, full undo/redo, and organized asset management.
• Enterprise Features: SynthID watermarking for AI provenance and IndexedDB for offline asset caching.

Maintenance & Community

The project is maintained by Mark Fulton. Community engagement and additional resources, including live build sessions and courses, are available through the "Vibe Coding is Life Skool Community."

Licensing & Compatibility

The project is licensed under the GNU Affero General Public License v3.0 (AGPL-3.0). This license permits free use for personal and commercial projects, but requires that any modifications or derivative works distributed or made available over a network must also be shared under the AGPL-3.0 license.

Limitations & Caveats

The current implementation uses direct client-side API calls; a backend proxy is recommended for production. It requires modern browsers with Canvas/WebGL support and is subject to Google AI Studio rate limits. While optimized for 1024x1024 outputs, Gemini model output dimensions may vary.