v1.0.0 - Initial Working Implementation

This is an early-stage experimental desktop image editing utility that leverages Google's Gemini Flash 2.5 Images (Nano Banana) for AI-powered image transformations.

✨ What's New

This initial release includes:

• Working Gemini API Integration - Reliable image-to-image transformations using Google's Gemini API
• Desktop GUI - Tab-based interface for managing multiple images and editing workflows
• Prompt Template Library - 14 categories of pre-written prompts for common image edits
• Custom Prompts - Support for writing your own transformation prompts
• Version Control - Automatic backup and versioning of edited images
• Multiple Aspect Ratios - Support for 1:1, 16:9, 21:9, and 9:16 formats
• Secure API Storage - Uses system keyring for secure API key storage

Installation

Quick Start

1. Clone the repository
2. Run ./setup-venv.sh to set up the environment
3. Run ./run.sh to launch the application

Building from Source

To create a standalone executable:

./build.sh

This creates a self-contained executable at ./dist/nano-banana-desktop

Features

Prompt Template Categories

• Artifact Removal
• Artistic Effects (comic book, impressionist, pop art, stained glass)
• Atmosphere (dramatic lighting, golden hour, vintage film)
• Avatars (cartoon, photorealistic)
• Background Editing
• Color Adjustments (B&W, saturation, sepia, vibrant)
• Enhancements (denoise, sharpen)
• Expression Edits
• For Fun (add animals, family photo edits, cultural contexts, historical periods)
• Graphics (diagrams, banners, favicons, stickers, wireframes)
• Style Transfer (oil painting, pencil sketch, watercolor)
• Subject Separation
• Text Editing

Future Plans

• Support for additional API providers (Claude, Replicate, etc.)
• Enhanced editing capabilities
• More sophisticated prompt management
• Performance optimizations
• Expanded template library

Important Notes

• Work in Progress: This is an experimental v1.0.0 release
• API Key Required: You'll need a Google Gemini API key
• Linux Only: Currently tested on Ubuntu with KDE/Qt support
• AI-Assisted Development: This project was built using Claude Code

Technical Stack

• GUI: PySide6
• API: google-genai
• Image Processing: Pillow
• Security: keyring

Documentation

See the README for full documentation.

Screenshots