⚡ PixelSpark

Professional AI Photo Editor for Android

AI-powered photo editing that runs entirely on your device

---

📸 Results Showcase

---

🌟 Features

• 100% Offline AI: Uses MIRNet TFLite to enhance low-light images entirely on device. No internet required.
• High-Performance C++ Core: Custom native processing pipeline using dart:ffi for maximum speed and efficiency.
• Non-Destructive Editing: Robust undo/redo history with memory-efficient JPEG compression.
• Pro Editing Suite: 21+ professional tools including AI enhancement, style filters, and manual adjustments.
• Premium UI/UX: Modern white and sky-blue aesthetic with smooth animations and responsive layout.

🛠 Tech Stack

Layer	Technologies Used
Frontend	Flutter, Provider, Google Fonts, Shimmer
Backend/Core	C++17, CMake, dart:ffi
Machine Learning	TensorFlow Lite (tflite_flutter), MIRNet
Architecture	Provider State Management, Native Background Isolates

🚀 Quick Start

1. Prerequisites

• Flutter SDK: >= 3.10.0
• Android SDK: Platform 35, NDK 26.1.10909125
• Java: JDK 21 (Android Studio bundled is recommended)

2. Environment Setup

The project includes automated scripts to handle dependencies and asset generation.

# Run development build on connected device
./scripts/run.sh

# Build production-ready APK
./scripts/build_apk.sh

🧰 The Pro Toolset

• AI Adjustments: Smart Enhance (MIRNet), Optimization Engine Fallback.
• Professional Filters: Pencil Art, Anime Style, Color Pop, Comic Book, 3D Relief, Warm Classic, Retro Film, Black & White.
• Manual Control: Brightness, Contrast, Saturation, Sharpness, Exposure, Gamma.
• Creative Retouch: Soft Focus, Smooth Skin, Noise Clean, Edge Art, Sketch Style.
• Geometry: Rotate, Flip, Free Crop.

🧠 Machine Learning Details

PixelSpark features an advanced MIRNet architecture optimized for mobile deployment.

• Model Size: ~27MB
• Processing: Parallel isolate-based inference to keep the UI fluid.
• Compatibility: Automatically detects hardware support and selects the optimal processing engine (AI vs Native C++).

🤝 Contributing

Contributions are welcome! Please ensure all UI changes adhere to the predefined AppTheme and heavy processing is offloaded to isolates via MLService or NativeProcessor.

📜 License

MIT License. See LICENSE for details.