Hello Everyone,
After much testing, I have found that PixInsight's new color calibration process (PCC) works quite well. I suggest always using the Average Spiral Galaxy (ASG) white reference. The PI G2V white reference produces much redder results and does not seem very useful. I tried some of the other white reference models and got wildly varying results. I really don't see why they are even offered. Maybe the developers can give us some further insight on this.
With ASG, the PCC process will show the color as seen from Earth. If there is intervening galactic extinction, galaxies will appear reddened, with correct foreground star color. This may be new ground for those who have been using the integrated light of galaxies for a white reference. Galaxies that should be reddened will no longer be blue. IC 342 is a great example. Here is a shot with PCC color.
http://bf-astro.com/pccTest/ic342_PCC.jpgWith the success of PCC, using the ASG white reference, I really don't see the need for using any of the other PixInsight color calibration procedures. However, if you want to show a reddened galaxy with its intrinsic color, you can still do so. But remember, you will have incorrect color in the foreground stars.
Now, just because I have to do it, here's how PCC compares with eXcalibrator (eXcal). eXcalibrator has two photometric calibration routines. The eXcalibrator Classic (white star) method uses stars, of any type, that should appear white. With galactic extinction, some slightly blue stars can appear white. Secondly, eXcalibrator has a linear regression routine. This uses stars that should appear slightly blue to white and to yellow/red. These two methods usually get similar and often identical results.
I processed 25 images that included galaxies, emission nebulae, clusters and reflection nebulae. The results were normalized to the red. PCC consistently had slightly bluer/cyan results. The two programs always agreed in the general direction of the color correction.
PCC (Average Spiral Galaxy) vs. eXcal (Linear Regression) PCC averaged 6.12% more blue
Std Dev 5.57
90% confidence 4.29% to 7.95%PCC averaged 1.28% more green
Std Dev 3.13
90% confidence 0.25% to 2.31%PCC (G2V) vs. eXcal (White Star) eXcal averaged 12.67% more blue
Std Dev 3.31
90% confidence 11.26% to 14.07%eXcal averaged 7.27% more green
Std Dev 3.13
90% confidence 5.94% to 8.60%With galaxies and clusters, the difference in the two programs was less than the average, with PCC consistently bluer. With images, dominated by nebula, the difference was greater than the average, with PCC still bluer. This may indicate a difference in eXcalibrator's use of Source Extractor aperture photometry and PixInsight's implementation. At the high-end of the 90% confidence range, emission nebula will very likely look a bit bluer with the PCC results.
A 4 to 8% change in color can be difficult to detect, especially with LRGB images. For some, it may require blinking the images. Others, with slightly diminished color vision, will not see the difference.
The following two images share the same luminance. With the eXcalibrator image, the blue was reduced by 6.12% and the green by 1.28%. They were processed identically with PI. The PCC image is noticeably bluer in the spiral arms. At the galaxy core, the eXcalibrator image is slightly redder.
Without the luminance data, these images show a slightly stronger difference. The PCC image has more faint detail in the blue arms. However, with galaxies, this is probably meaningless as most users create LRGB images.
http://bf-astro.com/pccTest/ngc5033_PCC.jpg
http://bf-astro.com/pccTest/ngc5033_eXcal.jpg My thoughts about using PCCIf you are working with an imported color FITS, the plate solve may fail. I found that splitting the RGB channels and reassembling the image solves the problem. This does not seem to make much sense, but it works.
I suggest starting with an image that has repaired gradients and a neutral flat background. Then let PCC do background neutralization after it corrects the color.
If you take many sub exposures and have strong data, I suggest setting the APASS magnitude limit to 16. This way PCC will use more stars. If your data are weak, you should probably leave "automatic limit magnitude" checked. Currently the APASS data are considered reliable to magnitude 16. In the future, they plan to reach magnitude 17.
Well, if you got this far you are very patient. I found the implementation of PCC to be excellent. I suspect we will start seeing fewer bright blue galaxy images.
Regards,
Bob
