Ugly star halos - what am I doing wrong?

[ldipenti]

Hello all,

I'm still trying to get better with StarTools, and I'm having a common issue with all my images: the resulting star halos look strange, and I think I'm not overprocessing the images, almost all modules I use default values, so here I am, asking for your opinion.
I'll use my latest capture, the Rosette Nebula, around 15 (13h bortle 9 + 2h bortle 3) hours of integration time, here's the stacked data:

https://siasky.net/EABogtOyYFBFh6S413hy ... zkhv6BTCtQ

And here's a preview of what I got with ST 1.7:

Rosette_Nebula-15hs.jpg (289.93 KiB) Viewed 4393 times

This is the zipped log file:

https://siasky.net/AABWzRdITIVw4UhD5ZHc ... p0o6n8t-9A

My usual way of using StarTools is going module after module from top to bottom, left to right as suggested in the documentation. Maybe what I'm doing wrong is not correcting star masks, or something like that.

Any suggestions will be appreciated, and you're more than welcome to give my data a try, I would love to see how much potential it has

Thanks in advance,
Lucas.

[admin]

Hi Lucas,

Would you be able to show a 100% zoom crop?

I'm seeing some aberrant color information in the highlights, which, when processed further, can cause fringing to become worse.
You should be able to tell when the problem starts appearing as you process. What step was that?

EDIT: I just now realised you are combining datasets from different conditions, which is not recommended an is likely the cause of the fringing (stars will be over-exposing differently, while noise levels will vary greatly as well).

Try stacking either the Bortle 3 or Bortle 9 data separately.

[ldipenti]

Hi Ivo!

Sorry, I shouldn't have sent the multi-bortle stack, but this is happening even when I use one set of shots.
I've now worked on the bortle 3 stack, uploaded the here if you want to take a look:

https://siasky.net/CABKiap5DMFPxZQvwzG4 ... 0iCie5qCtA (the FITS file)
https://siasky.net/AACERk7PvAQZJgbaPyXK ... E72WOuitlA (the resulting processed TIFF image)

A 100% crop of the resulting image would be:

100% Crop.jpg (332.61 KiB) Viewed 4361 times

My current equipment is:
* Askar FRA400/5.6 scope
* SkyWatcher AZ-Gti in EQ mode
* Nikon D500 DSLR (cropped sensor)
* ZWO ASI120MM mini guide cam
* ZWO ASIAIR Pro astro computer

This stack is almost 2h of integration time taken from the DSLR's RAW files, 300sec subexposures with dithering every other subexposure. I chose this over the bortle 9 to avoid over exposing stars as much as possible for this example, I think the histogram peak is around 25% from the left.
Stacking was made using AstroPixelProcessor with the suggested settings for StarTools, flats taken just after every shoot (from 2 consecutive nights) and scaled master dark (I made a master dark file with AAP of around 24hs integration time, 300 secs subexposures) and master bias files.

I've made a step by step image with 100% zoom, hope it's useful (had to crop a lot to get under 500kb)

step-by-step.jpg (481.35 KiB) Viewed 4361 times

Thanks!

[admin]

Thank you for uploading the Bortle 3 stack (could it be some sort of filter was used on that one by the way?)
As you can see in the crop below, there is still aberrant color information present (fringing and subtle channel mis-alignment).

Selection_491.jpg (137.87 KiB) Viewed 4204 times

It's slight, but will be enough to start bleeding into neighbouring pixels if used as a source by, for example, the Shrink module.
I'm not sure what the source might be - it could be your stacker having a hard time aligning this 100% or (less likely) some sort of feature of your optics. The problem appears severe enough for the default color calibration to be thrown off (emblematic of aberrant color information in the highlights such as stars).
In cases like these, a post-processing workaround would be to use one of the "Artistic" styles (for example the Legacy preset) in the Color module. This mimics the way less sophisticated software tends to desaturate highlights, which in the process, conveniently hides the problem. Even then some aberrant coloring can be seen furtherer away from the cores.
This is more than likely the cause/genesis of the halos. When using the Shrink module, however, you can try increasing the 'Color Taming' parameter, which will filter the chrominance signal (color signal) while Shrinking stars. You can also try running Shrink before you run the Color module (e.g. when your image is still mono).

Let me know if that works for you.

[ldipenti]

Hi Ivo,

I tried your suggestions without much improvement, so I tried stacking the best 20% & 50% of the data I have, and I think I got much better results, so my question is: Do you think focus discrepancies within an imaging session could produce the color fringing that was appearing?
I'm considering in getting an EAF because I just set the focus at session start time and leave the gear working all night and checking by eye I can see that sometimes the focus is off.
Now that we're about to start the cool season, this will surely exacerbate because temp differences will be greater through the night.

best regards

[admin]

Hi Ivo,

I tried your suggestions without much improvement, so I tried stacking the best 20% & 50% of the data I have, and I think I got much better results, so my question is: Do you think focus discrepancies within an imaging session could produce the color fringing that was appearing?

Absolutely! This can indeed very well explain exactly these sorts of problems in the highlights. This would constitute one of the more extreme cases, but, yes. Another cause could be significant changes in atmospheric conditions. And when both are at play, things can deteriorate quickly indeed.

I'm considering in getting an EAF because

If you leave your gear acquiring data unattended, then that may indeed make for a great addition.

Great sleuthing, and I'm glad you've found a likely culprit!

[ldipenti]

Hello Ivo,

Just for documenting reasons, I'll leave a comparison. And maybe you can clarify this: Is the FWHM measure a good way of evaluating the dataset in absolute terms? I mean: can I say for example, all subexposures above FWHM 4.5 should be discarded? It would be nice to have some numeric measure of the quality so that I could (re)evaluate my data (before integrating) through the years and be able to pick and choose whatever I'm aiming for on any given time.
I use AstroPixelProcessor for stacking and it seems that its scoring system isn't totally depending on FWHM.

Rosette comparison.jpg (432.9 KiB) Viewed 3891 times

[admin]

Just for documenting reasons, I'll leave a comparison.

It is much appreciated.

And maybe you can clarify this: Is the FWHM measure a good way of evaluating the dataset in absolute terms? I mean: can I say for example, all subexposures above FWHM 4.5 should be discarded? It would be nice to have some numeric measure of the quality so that I could (re)evaluate my data (before integrating) through the years and be able to pick and choose whatever I'm aiming for on any given time.

Unfortunately there are not measures you can use that work like that (reliably).
FWHM only says soemething about about an imaginary Gaussian profile fitted to a star, and how wide this profile is. It is mostly useful for estimating atmospheric turbulence (aka seeing) conditions, as light gets spread in a way that is somewhat close to a Gaussian profile.
For focus, however, light gets spread differently - more like concentric circles circles, rather than a Gaussian profile - so it's not really that great of a tool to use for measuring focus.