Retaining Faint Colors During Background Extraction and Color Calibration

wrichards

Well-known member
Jun 22, 2020
65
0
I have a stack of Rho Oph taken with a Canon 6D Mark II and an EF 135mm F/2L lens (60 Lights @ 60 sec, 60 Darks, 100 Bias, 32 Flats). After going through the registration and stacking process, I copped the edges a bit and tried to do background extraction.

Bear in mind that this image has an extremely dense star field just about everywhere, so there are precious few regions that are truly "background only" and they are very small. So that makes this a bit more challenging. (The edge of the Milky Way core starts to appear in the lower-left part of the image.)

I actually prefer to use DBE even though it's far more time consuming (to do it right), but DBE seems to remove a lot of the faint color in the nebulosity. ABE seems to be a bit less aggressive, so I used that on this stack. I always make sure I have 75-125 sample points and that none of them include any stars. In this case, I set the sample point box size to 10 pixels. Because of the dense star field in this image, placing points that contain "background only" is sometimes difficult so a small number of points had some extremely faint stars in them. I also placed no points anywhere near the nebulosity and it was impossible in the lower-left corner (in the Milky Way core). But when I look at what DBE is actually subtracting from the image, it's a much stronger red-orange image than what ABE extracts. Unfortunately, I didn't retain any of the images processed with DBE to show as a comparison but am curious if others experience the same result or if perhaps I am doing something wrong during the DBE process.

The second question I have is about PCC and (to a lesser extent) CC - they both appear to reduce faint reds. For some reason, my image doesn't have the strong red color that I've seen in other Rho Oph images (I don't have an Ha filter, but I've seen strong reds from other unmodded DLSRs), so I can't afford to lose any of the reds that I did manage to capture.

The first image below was only processed with ABE (and auto-stretched). The second image was processed with CC and SCNR to reduce the green tint that appears after running CC. Notice how you can still see some faint red nebulosity in the first image, but it's virtually gone in the second image. Also, the blues are noticeably stronger in the second image. So there appears to be a "blue shift" (pun intended). The full size images are HERE if the attachments are too small to see the effect. I tossed the PCC result because the effect was even worse using PCC.

So is this normal or do I need to change some settings?

The Rho Oph cloud complex should have a lot more red in it and I can't seem to bring much of it out after going through the background neutralization and color calibration processes.

Just in case anyone wants to play around with it, the unedited stacked image ("A) Rho Oph-DC.xisf") is also on the Google Drive site referenced above. The only thing I did was to crop a few rows of pixels around the edge to eliminate stacking artifacts.

A) Rho Oph-DC-ABE.jpg
A) Rho Oph-DC-ABE-PCC-SCNR.jpg
 

fredvanner

Well-known member
Apr 17, 2019
86
4
Wells, Somerset, UK
Could you post the approximate focal length and pixel size. (I could work it out by measuring the image and looking up the camera, but its easier and less error-prone if you post it). They would usually be in the header, but either they were never there, or they've been lost in ther processing. I need them if I'm going to look at PCC performance, since PCC needs to plate solve the image.
 

wrichards

Well-known member
Jun 22, 2020
65
0
Could you post the approximate focal length and pixel size. (I could work it out by measuring the image and looking up the camera, but its easier and less error-prone if you post it). They would usually be in the header, but either they were never there, or they've been lost in ther processing. I need them if I'm going to look at PCC performance, since PCC needs to plate solve the image.
Hi Fred, no problem.

Canon 6D Mark II
Pixel size: 5.67 um
Focal length: 135mm
Noise Level: 0.34
 

fredvanner

Well-known member
Apr 17, 2019
86
4
Wells, Somerset, UK
I find that running PCC on the unedited image makes almost no change - the image is already close to correct photometric white balance. DBE is going to struggle here, since there is almost no "background" in the frame. ABE also has problems, since there is lots of large scale structure that is not background. Also, some of the subtle background shades you are looking for are part of larger regions not in this frame, which makes it difficult to select and process them. It could be worth looking at the "PuWe 1 Processing Notes" item in the PI processing examples.