Focal Pointe Observatory
Astrophotography by Bob Franke

Recent Images
   Natural Color
   Narrow Band
Solar System
My Freeware
Tips & Tricks
Published Images
Local Weather


Send Email



































































































































Is It Necessary To Remove Color
Bias Before Applying Color Ratios?

Color bias and gradient removal are linear processes using only addition or subtraction. Color balancing uses multiplication or division. Therefore color bias and gradient removal are performed before color balancing. For this exercise, the gradients were removed from the individual red, green and blue images.

This test shows that removing the color bias before or after applying color ratios makes little or no difference in the final color. However, this is no excuse for not doing the processes in the proper order. It may make a difference.

Aperture photometry subtracts the surrounding background level from a star's brightness. So addition or subtraction pixel math has no effect on color balance calculations. When using several stars for determining color balance, gradient removal may have a very small, less than 1%, effect. The eXcalibrator results were nearly identical before and after gradient or color bias removal.


Software Used:

PixInsight (PI) for gradient removal.

eXcalibrator for color balance calculations.
CCDStack for pixel math and RGB image creation.


Changes caused by PixInsighrt scaling.

When PixInsight loads a FITS image, the data are immediately rescaled from 0.0 to 1.0. To use the images with eXcalibrator, the PI files were saved as 16-bit unsigned integer. With this format, PI uses a scale of 0 to 65,535. This may change the color balance. With the original data, the RGB color correction ratios are 1.000, 1.292 and 1.622. After PI rescaling the RGB ratios changed to 1.000, 1.255 and 1.631. EXcalibrator calculated the ratios by using 10 stars from the SDSS- DR9 data.

In this case the color balance change is minor. Since no data are lost, the rescaling is not a problem.




The first pair of images show the initial RGB color combines. Pixel math was used to remove the color bias. The blue data was increased 484 and the green reduced by 764. The pixel math was applied to unstretched images.


RGB Ratios 1,1,1 Before Color Bias Removal


RGB Ratios 1,1,1 After Color Bias Removal




The second two images are with the eXcalibrator color ratios applied to the above images. It is not surprising that a strong color bias is introduced.


RGB Ratios 1.00, 1.26, 1.63 applied w/o Color Bias Removal


RGB Ratios 1.00, 1.26, 1.63 Applied After Color Bias Removal




The final image pair is the above two images with the color bias removed using pixel math. These images are the answer to the original question.

The images are essentially identical.


Color Balance Set Before Removing Color Bias


Color Balance Set After Removing Color Bias