๐Ÿ“ธ GCC: Generative Color Constancy via Diffusing a Color Checker

๐ŸŒŸ GitHub Repository | ๐Ÿš€ Project Page | ๐Ÿ“„ Paper

Upload a RAW image to automatically correct its white balance using our diffusion-based color constancy method. GCC generates a color checker that accurately reflects the scene's ambient lighting, then uses it to estimate the illuminant and apply precise white balance correction.

Getting Started:

  1. Upload Your Image: Use the image upload box on the left to provide your RAW image (supports various bit depths).
  2. Process: Click the "๐Ÿš€ Process Image" button to start the color constancy correction.
  3. View Results: The generated color checker and white balanced result will appear below.
  4. Download: Right-click on any result image to save it as PNG to your device.
  5. Technical Details (Optional): After processing, you can view detailed processing information in the expandable section below.

๐Ÿ“ค Input Image

๐ŸŽฏ Generated Color Checker (512ร—512)

โš–๏ธ White Balanced Result

Processing Steps:

  1. Preprocessing: Resize input to 512ร—512, apply gamma correction (ฮณ=2.2)
  2. Color Checker Generation: Generate scene-harmonized color checker at image center using diffusion model
  3. Illuminant Estimation: Extract illuminant from gray patch (patch 20)
  4. White Balance Correction: Apply RGB gains in linear space, then apply gamma correction (ฮณ=2.2) for sRGB output

๐Ÿ“„ Citation

@InProceedings{Chang_2025_CVPR,
    author    = {Chang, Chen-Wei and Fan, Cheng-De and Chang, Chia-Che and Lo, Yi-Chen and Tseng, Yu-Chee and Huang, Jiun-Long and Liu, Yu-Lun},
    title     = {GCC: Generative Color Constancy via Diffusing a Color Checker},
    booktitle = {Proceedings of the Computer Vision and Pattern Recognition Conference (CVPR)},
    month     = {June},
    year      = {2025},
    pages     = {10868-10878}
}