StarTools

I don't to be the guy always posting processing challenges but I ran into another color processing challenge that I can't figure out after messing around with it all day yesterday. The problem is that StarTools is adding red rings around the stars in the image. For reference these stars should be blue based on other images online. This anomaly is repeated across the image. I did a process in photoshop and nebulosity and was unable to duplicate the issue. It appears that when StarTools is white balancing it is completely pushing blue from the other fringe. I tried to use a mask on the stars but was unable to resolve the issue.
In my attempt to resolve this issue I tried to stack in DSS using a couple of different methods that I normally don't mess with. One thing that I noticed is that even though I use a fairly high sigma of 3 in DSS it is still rejecting valuable data on my object itself. Below shows a comparison of average vs. Kappa Sigma Clipping with a Sigma of 3. You will note that the increase in signal on the object is offset by the increase in hot pixel noise as DSS lacks a bad pixel mapping feature.

First Kappa Sigma Clipping @ Sigma 3 and 5 iterations Now using average stacking

midwayexpress wrote:I don't to be the guy always posting processing challenges but I ran into another color processing challenge that I can't figure out after messing around with it all day yesterday. The problem is that StarTools is adding red rings around the stars in the image. For reference these stars should be blue based on other images online. This anomaly is repeated across the image. I did a process in photoshop and nebulosity and was unable to duplicate the issue. It appears that when StarTools is white balancing it is completely pushing blue from the other fringe. I tried to use a mask on the stars but was unable to resolve the issue.

First off, please keep posting challenges! It's the only way we learn and it grows the body of information on these forums. Asking questions is good!

StarTools recovers color where color information exists. Color information may have been introduced by incorrectly clipping the highlights (I've seen DSS do this before), introducing color where none existed previously. Looking at the stellar profiles, the red channel seems to be slightly misaligned and/or out of focus as well possibly contributing to the problem. Would you be able to upload the data somewhere so I can see what the issue is exactly? It'd be much appreciated!

If all fails, the Brightness Saturation response can be dialed back to mask the incorrect coloring in the highlights, but that's obviously a bit of a hack.

starissues.jpg

In my attempt to resolve this issue I tried to stack in DSS using a couple of different methods that I normally don't mess with. One thing that I noticed is that even though I use a fairly high sigma of 3 in DSS it is still rejecting valuable data on my object itself. Below shows a comparison of average vs. Kappa Sigma Clipping with a Sigma of 3. You will note that the increase in signal on the object is offset by the increase in hot pixel noise as DSS lacks a bad pixel mapping feature.

First Kappa Sigma Clipping @ Sigma 3 and 5 iterations

beforenoisestarissues.jpg

Now using average stacking

avg2Nforpost.jpg

At the posted resolution either stacks look decent (except for the obvious hot pixels issues). Kappa Sigma Clipping seemed to have done its job!
You could, of course, apply bias frames or dark frames - DSS should be able to use those to mitigate/eliminate the hot pixels.

I considered a red channel focus/alignment issue but I use a one shot color camera so I don't have much control over either. Its very possible that it is a deep sky stacker issue I'm looking at buying either MaximDL or CCDStack as they both seem to return a stack with similar FWHM. The problem with MaximDL is that its not quite as transparent in the registration process. CCDStack seems to produce nicer images but is poorly documented, has a poor debayer process, and the author is a big fan of processing each OSC channel separately which only would produce decent resolution if you had a long focal length setup that resulted in a large oversample. I also own nebulosity. Nebulosity is great at many things but the alignment process involves manual clicking on each image twice, writing literally 3 or 4 intermediate files, and the results have a FWHM of about .5" worse.

Anyways here are the stacked images to play with.

Average Stacking
http://1drv.ms/1BZO9Tu

Kappa Sigma Stacking
http://1drv.ms/1BZO5TM

Kappa Sigma Stack saved as 32 bit TIFF
http://1drv.ms/1QqiR3S

midwayexpress wrote:I considered a red channel focus/alignment issue but I use a one shot color camera so I don't have much control over either.

I was actually thinking chromatic aberration (e.g out of focus at some wavelenghts), but this usually manifests itself in the blue channel as well.

Its very possible that it is a deep sky stacker issue I'm looking at buying either MaximDL or CCDStack as they both seem to return a stack with similar FWHM. The problem with MaximDL is that its not quite as transparent in the registration process. CCDStack seems to produce nicer images but is poorly documented, has a poor debayer process, and the author is a big fan of processing each OSC channel separately which only would produce decent resolution if you had a long focal length setup that resulted in a large oversample. I also own nebulosity. Nebulosity is great at many things but the alignment process involves manual clicking on each image twice, writing literally 3 or 4 intermediate files, and the results have a FWHM of about .5" worse.

Be sure to consider PixInsight as well; the stacking and alignment modules are top notch!

Anyways here are the stacked images to play with.

Average Stacking
http://1drv.ms/1BZO9Tu

Kappa Sigma Stacking
http://1drv.ms/1BZO5TM

Kappa Sigma Stack saved as 32 bit TIFF
http://1drv.ms/1QqiR3S

Thanks for uploading. I had a look and the reddish color is unmistakeably there, even before Color calibration. Now that I'm able to have a closer look, chromatic aberration combined with mis-alignment of the color channels could be what's going on. There are clear fringes to even small stars... Do you know what's being used for the debayering stage?

How do your individual subs look, by comparison?

So I did a stack in the CCDStack demo which I have uploaded here:

http://1drv.ms/1f631KV

I think the stack is much better but the stars still have red fringing.

In terms of de-baying I tried both bilinear (which was used on both files I uploaded) and AHD in DSS. CCDStack doesn't specify the method that was used because, again I don't think the guy likes when people do that, but I suspect it is bilinear.

Good question on the individual frame. It looks like the stars are still a little fringed.

So here is the million dollar question, until M63 I really never had this problem... As I use a OSC I have no way of fixing a focus or alignment issue in one channel. What should I do?

So I figured this error out. It is related to the scientific colors setting. When scientific colors is turned to artistic detail aware the red rings disappear. See below.

That's unfortunately not really fixing the problem (it merely hides it by the inherent desaturation of brighter pixels of that method). If you look closely the problem is still there.
As I said, if all fails, the Brightness Saturation response can be dialed back to mask the incorrect coloring in the highlights, but that's obviously a bit of a hack (similar to turning Color Constancy off, but not as bad) .