ComfyUI Extension: ComfyUI Latent Color Tools

Authored by DenRakEiw

Created

Updated

0 stars

Advanced color manipulation and image adjustments directly in ComfyUI's latent space

Custom Nodes (0)

    README

    ComfyUI Latent Color Tools

    A collection of powerful nodes for color manipulation and image adjustments directly in the latent space, providing faster and more efficient processing compared to traditional image-based operations.

    🚀 Features

    🎨 Latent Color Match

    Advanced color matching between latents using multiple algorithms:

    • Cubiq-based methods with kornia color space conversions (LAB, YCbCr, LUV, YUV, XYZ, RGB)
    • Advanced algorithms using color-matcher library (hm-mkl-hm, mkl, hm, reinhard, mvgd, hm-mvgd-hm)
    • Real-time processing directly in latent space
    • Batch processing support for efficiency

    🎛️ Latent Image Adjust

    Complete image adjustment suite working in latent space:

    • Brightness (-1.0 to 1.0) - Additive brightness adjustment
    • Contrast (0.0 to 3.0) - Multiplicative contrast around mean
    • Hue (-180° to 180°) - Color tone shifting with HSV conversion
    • Saturation (0.0 to 3.0) - Color intensity adjustment
    • Sharpness (0.0 to 3.0) - Unsharp masking and blur effects

    📦 Installation

    Method 1: ComfyUI Manager (Recommended)

    1. Install ComfyUI Manager
    2. Search for "Latent Color Tools" in the manager
    3. Install and restart ComfyUI

    Method 2: Manual Installation

    1. Navigate to your ComfyUI custom nodes directory:

      cd ComfyUI/custom_nodes/

    2. Clone this repository:

      git clone https://github.com/yourusername/ComfyUI-Latent-Color-Tools.git

    3. Install dependencies:

      cd ComfyUI-Latent-Color-Tools
pip install -r requirements.txt

    4. Restart ComfyUI

    🎯 Usage

    Latent Color Match

    1. Add the "🎨 Latent Color Match" node to your workflow
    2. Connect your source and reference latents
    3. Choose a color matching method:
      • LAB: Best for natural color matching
      • hm-mkl-hm: Highest quality (requires color-matcher)
      • YCbCr: Good for skin tones
      • RGB: Direct channel matching
    4. Adjust the factor (0.0-3.0) to control the effect strength

    Latent Image Adjust

    1. Add the "🎛️ Latent Image Adjust" node to your workflow
    2. Connect your latent input
    3. Adjust parameters as needed:
      • Brightness: Negative values darken, positive brighten
      • Contrast: Values < 1.0 reduce contrast, > 1.0 increase
      • Hue: Shift color tone in degrees
      • Saturation: 0.0 = grayscale, > 1.0 = more vibrant
      • Sharpness: < 1.0 = blur, > 1.0 = sharpen

    🔧 Technical Details

    Dependencies

    • kornia (>= 0.6.0) - For advanced color space conversions
    • color-matcher (>= 0.2.0) - For professional color matching algorithms
    • torch - PyTorch (included with ComfyUI)
    • numpy - Numerical operations (included with ComfyUI)

    Performance Benefits

    • No VAE encoding/decoding - Works directly with latents
    • GPU accelerated - Full CUDA support
    • Batch processing - Efficient handling of multiple samples
    • Memory efficient - Lower VRAM usage compared to image operations

    Supported Tensor Shapes

    • 4D tensors: [batch, channels, height, width]
    • 5D tensors: [batch, channels, 1, height, width] (automatically handled)
    • Any number of channels (RGB-like processing for first 3 channels)

    📊 Comparison with Image-based Methods

    | Feature | Latent Space | Image Space | |---------|-------------|-------------| | Speed | ⚡ Fast | 🐌 Slow | | Memory Usage | 💾 Low | 📈 High | | Quality Loss | ✅ None | ❌ VAE artifacts | | Integration | 🔄 Seamless | 🔀 Requires conversion |

    🎨 Color Matching Methods

    Cubiq-based (with kornia)

    • LAB: Perceptually uniform color space, best for natural images
    • YCbCr: Separates luminance from chrominance, good for skin tones
    • LUV: Alternative perceptual color space
    • YUV: Broadcast standard color space
    • XYZ: CIE standard color space
    • RGB: Direct RGB channel matching

    Advanced (with color-matcher)

    • hm-mkl-hm: Histogram + Monge-Kantorovich + Histogram (highest quality)
    • mkl: Monge-Kantorovich Linearization
    • hm: Classical Histogram Matching
    • reinhard: Reinhard et al. method
    • mvgd: Multi-Variate Gaussian Distribution
    • hm-mvgd-hm: HM + MVGD + HM compound

    🐛 Troubleshooting

    Common Issues

    "Kornia not available" warning

    pip install kornia>=0.6.0

    "Color-matcher not available" warning

    pip install color-matcher>=0.2.0

    Tensor shape errors

    • The nodes automatically handle 4D and 5D tensors
    • If you encounter shape issues, check your latent source

    Weak effects

    • Increase the factor parameter (try 1.5-3.0)
    • Some methods work better with specific content types

    📝 Changelog

    v1.0.0

    • Initial release
    • Latent Color Match with multiple algorithms
    • Latent Image Adjust with 5 adjustment types
    • Full kornia and color-matcher integration
    • Automatic tensor shape handling

    🤝 Contributing

    Contributions are welcome! Please feel free to submit a Pull Request.

    📄 License

    This project is licensed under the MIT License - see the LICENSE file for details.

    🙏 Acknowledgments

    • cubiq for the original ImageColorMatch implementation
    • kornia team for excellent computer vision library
    • color-matcher for professional color matching algorithms
    • ComfyUI community for the amazing platform

    📞 Support

    If you encounter any issues or have questions:

    1. Check the Issues page
    2. Create a new issue with detailed description
    3. Include your ComfyUI version and error logs

    Made with ❤️ for the ComfyUI community