🎮 Pixel-PFP 🕹️ Generate retro-styled 8-bit profile pictures from text prompts!
Pixel-PFP uses AI to generate amazing retro-styled profile pictures from simple text prompts:
- 🤖 AI-Powered Generation: Create unique images using StableDiffusion models
- 🎨 8-bit Style Conversion: Multiple conversion methods including K-means clustering + Floyd-Steinberg dithering
- 🚀 Dual Backends: Choose between PyTorch or Rust-based Candle framework
- ⚡ Optimized Performance: Rust implementation offers 3-5x speed boost
- 🎭 Customizable Effects: Control palette size (4-256 colors), pixel size, and dithering intensity
Coming soon - placeholder for example images grid
Prerequisites:
- Python 3.8+
- Rust 1.65+ (for Candle backend)
# Python dependencies
pip install torch diffusers pillow scikit-learn
# Rust components (optional)
cargo install --path candle-componentsBasic Command:
python generate_8bit_pfp.py "your fav anime character" Options:
| Flag | Description | Default |
|---|---|---|
--prompt |
Text prompt for generation | Required |
--method |
Conversion method (basic/pyxelate/rust) | rust |
--palette-size |
Colors in final image (4-256) | 8 |
--pixel-size |
Pixel block size | 8 |
--dithering |
Enable dithering (true/false) | true |
--output |
Output path | output.png |
In Progress
✅ Basic 8-bit conversion pipeline
✅ Rust backend implementation
🟧 [ ] 8-bit quantized model support
🟧 [ ] 4-bit quantized model optimization
🟧 [ ] New Rust pixelation engine (replacing Pyxelate)
🟧 [ ] Classic console palettes (NES, Game Boy, C64)
🟧 [ ] Batch processing mode
Future Enhancements
[ ] Web-based interface
[ ] Animated pixel art generation
[ ] Cross-platform desktop app
[ ] Social media integration
[ ] NFT export capabilities
Core Components:
- Image Generation
- PyTorch:
diffusersStable Diffusion pipeline - Rust: Candle framework with LORA adapters
- PyTorch:
Pixelation Engine
| Method | Language | Features | Speed |
|---|---|---|---|
| Basic | Python | K-means + Floyd-Steinberg | 1x |
| Pyxelate | Python | Advanced dithering | 0.8x |
| Rust | Rust | SIMD optimized pipelines | 3-5x |
We welcome contributions! Priority areas:
- Performance optimizations
- New pixelation algorithms
- Additional model support
- Web UI development
- Documentation improvements
MIT License - See LICENSE for details
- Stability AI for Stable Diffusion models
- Pyxelate by @dandrino
- Candle ML framework
- Open source contributors everywhere!
Made with ❤️ and a lot of pixels