Page 1 of 1

1.8 SVDecon (post1)

Posted: Mon Jan 24, 2022 10:25 am
by Heno
Hi
I have literally spent hours with this module trying to figure out what I'm doing wrong. In the end I don't think I am. SVDecon is doing a great job with the nebula, but it is ruining my stars. SVDecon pulls a lot of light into the star center. I will illustrate the problem with examples.
I have tried a wide range of settings without any furter success. Since I can only attach 3 files in a post, I have split this topic in in two posts.
I have followed all the normal steps down to SVDecon.

Re: 1.8 SVDecon (post1)

Posted: Tue Jan 25, 2022 5:24 am
by admin
Heno wrote: Mon Jan 24, 2022 10:25 am but it is ruining my stars. SVDecon pulls a lot of light into the star center. I will illustrate the problem
Hi,

Pulling light back into point lights is precisely the purpose of deconvolution.
Your stars are, as far as I can tell, improved and better defined. If deconvolution works for the rest of the nebulosity, then it works for the stars as well.

If light is "re-concentrated" beyond unity (e.g. beyond the "whitest white" the current stretch allows), you can re-allocate dynamic range, to make head room for the whiter-than-white highlights either by;
  • trying the "Dynamic Range Extension" parameter
or
  • accepting the Decon result and re-stretching the linear (but now deconvolved) data with a third AutoDev
Does that help?

Re: 1.8 SVDecon (post1)

Posted: Wed Jan 26, 2022 9:48 am
by Heno
Thanks for your reply.
admin wrote: Tue Jan 25, 2022 5:24 am Pulling light back into point lights is precisely the purpose of deconvolution.
Your stars are, as far as I can tell, improved and better defined. If deconvolution works for the rest of the nebulosity, then it works for the stars as well.
I was afraid you would say that, and I'm sure it is theoretically correct, but still I don't like it and I honestly think my stars looks better when not deconvoluted. I will certainly try your proposals and see if it will improve things.
I would have loved to have the option of setting an inverted star mask (or any mask for that matter) before using the result from the selected stars so that stars (in this case) could be omitted by choice.

Re: 1.8 SVDecon (post1)

Posted: Wed Jan 26, 2022 10:26 am
by Stefan B
You could undo the stars' decon after the fact by applying a star mask, entering the Layer module and use the 'Undo Fg' or 'Undo Bg' option. The non masked part of the image should still be deconvoluted while the stars should appear rather blurry.

Regards
Stefan

Re: 1.8 SVDecon (post1)

Posted: Fri Jan 20, 2023 7:24 pm
by hixx
Hi Helge,
I am not sure if I understand what You would define as "look better". A matter of taste might be worthless to discuss, but probably one needs to get used to it.
Please look at the vertical triple star towards the bottom left side of your image. Without Decon it is more like a vertical stripe, the second image clearly presents 3 individual stars. Higher Deringing in image 3 again takes away deconvolution effect
Now, assuming we would be comparing optics here, it would be fair to say, image 2 "resolves better" than image 1. That is exactly the kind of improvement Decon gives You: Presenting stars as if view from high altitude under improved seeing condition...
Clear Skies, Jochen