ComfyUI Node: Color Correction Viewer (beta)

Color Correction Viewer (beta):

The ColorCorrectionViewer node is designed to enhance the visual quality of images by correcting color shifts that occur during the VAE (Variational Autoencoder) encoding and decoding process. This node is particularly beneficial for AI artists who work with generative models, as it ensures that the colors in the final output image closely match the original input image. By offering multiple correction methods, such as advanced 3D LUT, luminance zones, histogram matching, and statistical matching, the node provides flexibility and precision in color correction. The main goal of the ColorCorrectionViewer is to maintain the artistic intent of the original image while compensating for any color distortions introduced by the VAE, thus preserving the integrity and aesthetic quality of the artwork.

Color Correction Viewer (beta) Input Parameters:

original_image

This parameter represents the original input image before it undergoes VAE encoding. It serves as the reference for color correction, ensuring that the final output image retains the original color characteristics. The input type is IMAGE.

processed_image

This parameter is the image after VAE decoding, which may have color shifts. It is the target for color correction, where adjustments are made to align its colors with those of the original image. The input type is IMAGE.

correction_strength

This parameter controls the intensity of the color correction applied to the processed image. It ranges from 0.0 to 1.0, where 0.0 means no correction and 1.0 means full correction. The default value is 0.8, allowing for a balanced correction that effectively mitigates color shifts without overcompensating.

method

This parameter specifies the color correction method to be used. Options include luminance_zones, histogram_matching, statistical_matching, and advanced_3d_lut. Each method offers a different approach to color correction, with advanced_3d_lut being the most accurate for 3D color mapping. The default method is advanced_3d_lut.

preserve_inpainted

This boolean parameter determines whether to preserve new inpainted content in the processed image. When set to True, only areas that existed in the original image are corrected, leaving newly inpainted areas unchanged. The default value is True.

Color Correction Viewer (beta) Output Parameters:

corrected_image

The corrected image is the output of the node, where color shifts in the processed image have been adjusted to match the original image. This output ensures that the final image maintains the intended color palette and visual quality, making it suitable for artistic and professional use.

Color Correction Viewer (beta) Usage Tips:

• To achieve the most accurate color correction, use the advanced_3d_lut method, especially when working with complex color mappings.
• Adjust the correction_strength parameter to fine-tune the level of correction applied, starting with the default value of 0.8 and modifying as needed based on the specific requirements of your project.

Color Correction Viewer (beta) Common Errors and Solutions:

"Strength {adjusted_strength:.2f} > 1.0, clamping to 1.0 to prevent artifacts"

• Explanation: This message indicates that the calculated correction strength exceeds the maximum allowable value, which could lead to quantization artifacts in the image.
• Solution: The node automatically clamps the strength to 1.0 to prevent artifacts. You can manually adjust the correction_strength parameter to a lower value if needed.

"3D LUT correction failed: {e}, falling back to zone matching"

• Explanation: This error occurs when the advanced 3D LUT correction method encounters an issue, possibly due to insufficient data or computational limitations.
• Solution: The node automatically falls back to a simpler zone-based matching method. Ensure that your input images are of sufficient quality and size, and consider using a different correction method if the problem persists.