Author Topic: Is there a Pixinsight guru who would like to take a stab at processing my files?  (Read 1016 times)

Offline rtemen

  • Member
  • *
  • Posts: 56
    • View Profile
I was wondering if there is a guru out there who would like to process my files and log the steps to help me learn how to better process files myself.
I have been following several tutorials, but of course, none of them can tell about my files exactly.
The latest tutorial is from Kayron Mercieca on Light Vortex.
This made a big difference in my results, but there is still an issue with ringing on some of my stars.
Upon doing some more research, it is suggested that I use the DeConvolution process to help with this problem.
So, now I don't know what to do.
When should I have used the DeConvolution, and how do I tell that it is doing something positive?
I have saved my project, and I assume that I should be able to backtrack in my history and inject the DeConvolution Process???
Or do I just need to start over.

Kind of a long message, but thank anyone who might be willing to help.


Offline GJL

  • PixInsight Addict
  • ***
  • Posts: 102
    • View Profile
Rich, here is a tutorial about Deconvolution


Offline msmythers

  • PTeam Member
  • PixInsight Jedi
  • *****
  • Posts: 1176
    • View Profile
    • astrobin

Here is a posting by Juan from earlier this year. I think it states very clearly his views on the use of deconvolution.


Offline pfile

  • PTeam Member
  • PixInsight Jedi Grand Master
  • ********
  • Posts: 4471
    • View Profile
i don't think i was saying to use deconvolution, i was saying that the use of deconvolution or sharpening is a potential time where ringing around stars creeps in. if you did no decon or sharpening then you will have to go step by step thru your processing and see exactly where the ringing around the stars happens.


Offline the Elf

  • Newcomer
  • Posts: 16
    • View Profile

the first steps of processing (preprocessing) is something you can do like follow a step by step tutorial. There are very few decisions to make, like know if you have OSC or not and how many images of each kind to select the best pixel rejection. But sometimes there are parameters you have to adjust with quite a simple way of juding the result like the sigma for rejection. I recommend W.A. Kellers book for this.

Then comes a point where you have your master light and need experience. You have to decide if you do a certain step now, if you just forget it or if you attack a specific problem later. This applies to noise reduction, deconvolution and sharpening in my eyes. Lets talk about deconvolution, which was your initial question. Not being a guru but related to image processing in my job I'd like to say this: If your convolution comes from your optics only because you have no tracking issues and if you have plenty of signal and if the stars are already small the "Deconvolution" process in linear state can improve a good image even more. If you have fat stars or not circular stars and if this is different in the subs due to guiding problems, you should not do it. If you have a lot of noise and faint nebula you want to dig out of that noise you might be more satisfied with a slightlig blurred image with little noise instead. This means: denoise, go nonlinear, try ExponentialTransform with PIP and use smoothing and later build a star mask and use erosion in MorphologicalTransformation. Here is an example of that workflow:
That night the air was full of sand that originated in the sahara dessert and came down on central Europe and my tracking is not the best and due to lack of time only fery few lights were available. I tried deconvolution but it did not make any improvement. I tried to sharpen but the noise is to strong. So I live with this very smooth image with the noise corrected aggressivly which was the better alternative in my eyes. I'll try that object again next year. In the few projects I made so far there was only one that gained a small benefit from deconvolution (NGC6946 2017 in my CN gallery, but they reduced the resolution so much that you can't see the effect there). In all other cases it was better not to use it and reduce star size later.

As soon as you are nonlinear the technical guru needs to step aside and the artist has find his own interpretation of the scene. Nobody knows what it "really" looks like, because non of us will ever travel there and see it without our atmosphere and most of us do not have a scope with enough aperture to see the colors thru the eyepiece. So you have to do interpretations now. Rather than asking others to process your data you should look at others images first and decide if you like that style or not. I have several books with large high quality prints of popular space objects. I compare them all and decide which one is closest to my taste and I read what equippment was used and understand that my result will not reach that level. (If I become rich one day I'll buy the Hubble Space Telescope when NASA no longer needs it. This is the first day I'll use drizzle and deconvolution and I know it benefits.)

Anyway: send me a privat message with a link to your master light. If the rain goes on the whole weekend I might give it a try. May well be you are very satisfied with _your_ result after that.

the Elf

Offline the Elf

  • Newcomer
  • Posts: 16
    • View Profile
here is a crop including a 4x magnified view of some stars. I'd say if your stars are in that range (3-5 pixels diameter) and you have more than 3h exposure time like in this example it's worth trying Deconvolution. There are many images of this object in the web showing for more detail than mine. This is the best I ever out out of my AVX mount.