ImageEdit

An AI-powered image editing application with iOS and web interfaces that uses natural language prompts to transform images.

Features

• 🎨 Natural Language Editing: Describe changes in plain English
• 🖼️ Multi-Image Support: Combine elements from up to 3 images in a single edit
• 📱 iOS App: Native SwiftUI application with full Photos integration
• 🌐 Web App: Browser-based interface accessible from any device
• 🤖 Powered by Qwen-Image-Edit-2509: State-of-the-art AI model for image transformation
• 📊 Real-time Progress: Monitor generation status and progress
• 💾 Smart Memory Management: Automatic model loading/unloading to optimize GPU usage
• 🕐 History Tracking: Keep track of all your edits (iOS only)

Quick Start

Server Setup

1. Install Python dependencies:

   cd server
   pip install -r requirements.txt

2. Start the server:

   uvicorn image-edit-server:app --host 0.0.0.0 --port 8000

3. Access the web app: Open your browser to http://localhost:8000

iOS App

1. Open ImageEdit.xcodeproj in Xcode
2. Build and run on simulator or device
3. Configure server endpoint in Settings if needed

Project Structure

ImageEdit/
├── ImageEdit/              # iOS app source code
│   ├── ContentView.swift   # Main UI
│   ├── HistoryManager.swift # Edit history
│   └── Settings.swift      # User preferences
├── server/                 # Python backend
│   ├── image-edit-server.py # FastAPI server
│   ├── static/            # Web app files
│   │   ├── index.html     # Web interface
│   │   ├── app.js         # JavaScript logic
│   │   └── style.css      # Styling
│   └── requirements.txt   # Python dependencies
└── docs/                  # Documentation
    ├── architecture.md    # System design
    ├── api.md            # API reference
    └── web-app.md        # Web app guide

System Requirements

Server

• Python 3.8+
• CUDA-capable GPU (8GB+ VRAM recommended)
• 16GB+ system RAM

iOS App

• iOS 18.5+
• iPhone or iPad
• Xcode 15+ for development

Web App

• Modern web browser (Chrome, Safari, Firefox, Edge)
• JavaScript enabled

Key Features

Model Management

• Auto-load: Model loads on first request
• Auto-unload: Frees GPU memory after 30 minutes of inactivity
• Manual control: Unload button in both iOS and web interfaces
• Status monitoring: Real-time model status display

Image Processing

• Supports common formats (JPEG, PNG)
• Configurable generation parameters
• Progress tracking with percentage updates
• Queue system for multiple requests

API Endpoints

• GET / - Web application
• GET /health - Server and model status
• POST /edit - Submit image for editing
• GET /jobs/{id} - Check job status
• GET /results/{filename} - Download result
• POST /model/unload - Manually unload model

Environment Variables

• MODEL_TIMEOUT_MINUTES - Minutes before auto-unload (default: 30)

Development

See CLAUDE.md for AI assistant development guidelines.

Documentation

• Multi-Image Editing Guide - New! Learn how to combine multiple images
• Architecture Overview
• API Documentation
• Web App Guide
• Getting Started

License

MIT