Author Topic: ArcsinhStretch and White Center in Stars  (Read 974 times)

Offline GeneralT001

  • Member
  • *
  • Posts: 61
    • View Profile
ArcsinhStretch and White Center in Stars
« on: 2018 February 28 09:21:41 »
How do you avoid getting the white star center when using arcsinhStretch?

Only seems to occur on really big/bright stars.


Offline aworonow

  • PixInsight Addict
  • ***
  • Posts: 245
    • View Profile
    • Faint Light Photography
Re: ArcsinhStretch and White Center in Stars
« Reply #1 on: 2018 February 28 12:18:47 »
Problem likely results from over-stretching the stars such that the center saturates in all colors RGB). Try using a star mask.
Alex

Offline topboxman

  • PixInsight Addict
  • ***
  • Posts: 292
    • View Profile
    • Peter's Gallery
Re: ArcsinhStretch and White Center in Stars
« Reply #2 on: 2018 February 28 13:09:53 »
I too have started using ArcSinh and found it an excellent tool for stretching without harming the colors. This tool is a little sensitive and seem to prefer good data to begin with. In your case, it's possible that one of the three filter (RGB) image may have larger FWHM than other two filters and if that's the case, it seem to create large halo when stretching with ArcSinh.

Ever since I found out about ArcSinh about two weeks ago, I re-processed many of my old images with great success. There were a few that had unequal FWHM between Red, Green and Blue images and created funky looking stars when using ArcSinh.

I find using Star Mask making it worse because the halos will continue to grow outside of of stars of Star Mask.

Peter

Offline GeneralT001

  • Member
  • *
  • Posts: 61
    • View Profile
Re: ArcsinhStretch and White Center in Stars
« Reply #3 on: 2018 February 28 16:26:50 »
Thanks for the replies. I will check the FWHM for the RGB and go from there.

Cheers

Offline dpaul

  • PixInsight Addict
  • ***
  • Posts: 124
    • View Profile
Re: ArcsinhStretch and White Center in Stars
« Reply #4 on: 2018 February 28 17:38:14 »
What a fantastic tool - just tried it (never realised it existed).
Correct me if I'm wrong but it seems to do the same as Histogram Transfer and Curves Transform in one tool.

I've always stretched the data and then played around with darkening the background with Curve transformation (and an inverted mask to protect a galaxy).
Now with ArcsinhStretch I can stretch the data and darken the background much more effectively and with better colors. I presume its then no longer linear?

A question - what about the Lum content?  If I just use the process on the linear RGB and then add the lum, its washes out the result somewhat. On the other hand it seems less effective when I do use it on a linear LRGB image - which way is best?

Thanks

David

Offline topboxman

  • PixInsight Addict
  • ***
  • Posts: 292
    • View Profile
    • Peter's Gallery
Re: ArcsinhStretch and White Center in Stars
« Reply #5 on: 2018 February 28 20:47:45 »
I looked back my recently processed images using ArcSinh Stretch and it seems like the orange/yellow stars show halos around and whitish stars at the center. It might be a minor issue with ArcSinh tool for these orange/yellow stars. This is one example:

https://peternagy.smugmug.com/Telescopes/Galaxies/i-43cGvjP/A

This one has quite a bit larger FWHM from red filter images than green/blue filter images which is why stars look too reddish/pinkish and bright stars show large red/pink halos and white center:

https://peternagy.smugmug.com/Telescopes/Galaxies/i-P4wdzSq/A

Peter

Offline ChoJin

  • PixInsight Addict
  • ***
  • Posts: 106
    • View Profile
Re: ArcsinhStretch and White Center in Stars
« Reply #6 on: 2018 March 01 00:04:25 »
I get white centers too on my image i'm currently processing, but only with the version I have applied deconvolution to the L channel. With my data the stars looks fantastic when stretching the data without prior deconvolution.

I'm going to try doing the stretch in multiple passes (with a mild first stretch, and non-linear stretch afterward, possibly with a mask on those later stretches).

Otherwise I guess I'll go with the PixelMath-way, to combine the "with deconvolution" and "without deconvolution" versions.

I'll see how it goes...

Offline aworonow

  • PixInsight Addict
  • ***
  • Posts: 245
    • View Profile
    • Faint Light Photography
Re: ArcsinhStretch and White Center in Stars
« Reply #7 on: 2018 March 04 14:45:42 »
White is the saturation color: equal amounts of R, G, and B all sub-equal to 1. (Equal amounts less than saturation is some shade of gray.) There are at least three ways to avoid this: 1) do not stretch so aggressively. 2) Protect stars from saturation with an adequate mask--that may take some trial-and-error experimenting to find. It is not likely to be the default mask. And even with a good mask, you may need to dial back on the stretch. 3) Make a mask that reveals only the object you want to stretch and leaves everything else strongly or moderately masked. Linked below is a recent image that used a pretty aggressive arcsine stretch with a black mask everywhere but on the target. The stars were stretched separately and less aggressively with a star mask targeting the "larger" stars. The background was masked for both stretchings. My first attempts at arcsine stretching gave me the white centers too. Keep experimenting! good luck!

https://www.astrobin.com/full/335823/0/

Alex

Offline pfile

  • PTeam Member
  • PixInsight Jedi Grand Master
  • ********
  • Posts: 4389
    • View Profile
Re: ArcsinhStretch and White Center in Stars
« Reply #8 on: 2018 March 04 15:26:21 »
as ChoJin has noted, the deconvolution step tends to push energy to the center of the star, sometimes saturating them - to see this turn the STF off and then undo/redo the deconvolution. you'll see the bright stars go from more gaussian shapes to white discs at the core. so it helps to mask just the star cores while doing deconvolution.

rob

Offline ChoJin

  • PixInsight Addict
  • ***
  • Posts: 106
    • View Profile
Re: ArcsinhStretch and White Center in Stars
« Reply #9 on: 2018 March 05 02:29:37 »
this is what I eventually did: I masked the inner core of the stars during deconvolution and that fixed it. I used STF with modified sliders to focus on highlights when experimenting with my core mask.