Jump to content

NLCbanner2024.jpg.2478be509670e60c2d6efd04834b8b47.jpg

PI Initial processing question


Recommended Posts

Hi,

Having lost my astro mojo over the summer I've rekindled my interest over the last few days (helped by some clear nights!). It was unfortunate timing as I was trying to get to grips with PI at the time and I'm sort of starting from a basic level again now!

I'm trying to process a galaxy image I did over 3 nights in spring, M106 and for ease of learning I've stacked it using APP as I'm familiar with that but without selecting 'Neutralise background'. I then pulled the resulting FITS file in to PI. This has a very blue cast to it which is giving me problems being able to see what I'm doing with DBE. Is there any down side to running background neutralisation before DBE in PI? Or would I be better off stacking again in APP and selecting background Neutralisation this time?

Image below to demonstrate just how blue it is!

Thanks
Ed

M106.jpg

  • Like 1
Link to comment
Share on other sites

2 hours ago, edarter said:

Hi,

Having lost my astro mojo over the summer I've rekindled my interest over the last few days (helped by some clear nights!). It was unfortunate timing as I was trying to get to grips with PI at the time and I'm sort of starting from a basic level again now!

I'm trying to process a galaxy image I did over 3 nights in spring, M106 and for ease of learning I've stacked it using APP as I'm familiar with that but without selecting 'Neutralise background'. I then pulled the resulting FITS file in to PI. This has a very blue cast to it which is giving me problems being able to see what I'm doing with DBE. Is there any down side to running background neutralisation before DBE in PI? Or would I be better off stacking again in APP and selecting background Neutralisation this time?

Image below to demonstrate just how blue it is!

Thanks
Ed

M106.jpg

Re-stretch the image using the Screen Transfer Function but this time click on the Link RGB Channels icon to unlink RGB Channels. The background will then look neutral. The Link RGB Channels button is two buttons above the “nuclear” button. 
74F2CA55-8B43-4256-BC79-5D1DFD3AE05C.jpeg.0ee3bc1e51ca5c4df2d3c783655ffbd0.jpeg

Edited by Ouroboros
  • Like 1
  • Thanks 1
Link to comment
Share on other sites

1 hour ago, Ouroboros said:

Re-stretch the image using the Screen Transfer Function but this time click on the Link RGB Channels icon to unlink RGB Channels. The background will then look neutral. The Link RGB Channels button is two buttons above the “nuclear” button. 
74F2CA55-8B43-4256-BC79-5D1DFD3AE05C.jpeg.0ee3bc1e51ca5c4df2d3c783655ffbd0.jpeg

Ahhh thats better! Thank you! Lots of gradients etc to sort now but at least I can see what I'm doing in DBE now.

Steve, still stacking in APP as I'm more familiar with that. I'm at a point where I need a bit of a morale boost with my processing skills so need a bit of a quick fix on that side from PI. I really like the stacking in APP, but no so much the processing element. I will have a go at stacking in PI but as I've not been getting the results out of APP post processing that I want I would like to get that nailed in PI first.

Thanks
Ed

  • Like 1
Link to comment
Share on other sites

7 hours ago, edarter said:

Ahhh thats better! Thank you! Lots of gradients etc to sort now but at least I can see what I'm doing in DBE now.

Good.  Hopefully DBE will largely get rid of the blue background.  After DBE I re-stretch with RGB channels relinked so I can see an image more like what it really looks like. Keep us posted how it goes. :) 

  • Like 1
  • Thanks 1
Link to comment
Share on other sites

ok, so the unlinking worked great, but I'm still having trouble using DBE to get a nice even background. Something I always seem to struggle with! Can someone run DBE on the attached for me and show me what the result looks like? Its still blotchy for me to some extent no matter what I try with DBE. I've clearly not grasped one of the settings correctly yet!

Thanks
Ed

M106-session_1_session_2_session_3.fits

Link to comment
Share on other sites

I too often struggle with DBE, sometimes I can get it to work a treat yet on other images I struggle.

One thing I would do before DBE is crop the original to get rid of the artefacts due to stacking around the edges. This is the original with an auto stretch.
image.png.7c002077b9e72c5b3cd78daa7a2dae6b.png
Sometimes I get better results (or maybe I just find it easier) using ABE instead, just try ABE even using default parameters works reasonably well on your image, once using Division and then again using Subtraction.
This is how it comes out:
image.png.d6ab3e4c942b31d0b271b8480b28c0d9.png

Could be better but it is better than original
 

I have quickly tried with DBE and not been able to give it much effort as I am busy at moment but this is how I did applying DBE 3 times:
image.png.45df50407a40e15556e485e438159d11.png

Some areas of that better than the ABE version and some not as good but I think if you spend a bit of time carefully placing the boxes and maybe applying DBE a few times you will get there.
If nobody else shows us bth how its done by tonight I mat give it another go when I have finished my jobs for the day 🙂 

Sometimes I find it easier if you do a boosted auto stretch on the original shows you where to place the boxes a bit easier (click the radioactive button whilst holding shift down).

Steve

Edited by teoria_del_big_bang
  • Thanks 1
Link to comment
Share on other sites

8 hours ago, edarter said:

ok, so the unlinking worked great, but I'm still having trouble using DBE to get a nice even background. Something I always seem to struggle with! Can someone run DBE on the attached for me and show me what the result looks like? Its still blotchy for me to some extent no matter what I try with DBE. I've clearly not grasped one of the settings correctly yet!

Thanks
Ed

M106-session_1_session_2_session_3.fits 220.35 MB · 6 downloads

I had a bit of a go with this in pixinsight.  I notice re. the problematic  background  that there is actually a tiny amount of signal difference between the light and dark regions. Once I'd DBE'd it the difference between light and dark regions was something like 0.0020 AND 0.0024 in the RGB channels. I was hoping to back the background off by raising the black point.  It's a tad problematic though because some of the nebula's nebulosity is only just above the background.  nevertheless here's my very quick  attempt at processing your image.  It's not bad data. I think a bit more care with curves could get rid of the pesky background whilst retaining the nebulosity. 

Workflow:

Dynamic Crop
DBE
Colour Calibration.  I'd have done Photometric CC but the image data has been lost.
EZDenoise
SCNR  Taking out some green tint.
Histogram Transformation to nonlinear and to raise the black point  a bit
Curves Transformation to suppress background a tad and boost galaxy etc. 
Colour Saturation. I masked the background a bit to prevent colouring it. 

I didn't do any fancy stuff to pull out dynamic range detail in the nebula. 
 

M106_DBE_cc_ezdenoise_scnr_ht_ct_cs.thumb.jpg.eae68bb25e51fc01723525ab1ebfd9f1.jpg

 

 

  • Like 1
  • Thanks 1
Link to comment
Share on other sites

I'm wondering whether the flats were either not included in the stack or haven't worked correctly.

I trimmed off the excess from the stacking and then ran DBE and this is what it looks like with a STF Boost applied:

DBE.png.fac07f8d952b84020192e0dbd5fd733d.png

Notice the possible dust bunny in the top right, which the flats should have removed.

Then I put the above through GraXpert, which did a better job of removing the gradients, but there are still some left in there. Again, this is STF Boosted:

GraXpert.png.45771bd62866f1f215e36a6dfcaf31f1.png

Finally, I ran a full process to see what I could get.

My workflow was:

  • Dynamic Crop
  • DBE
  • GraXpert
  • BackgroundNeutralisation
  • IntegerResample (Bin x2)
  • HistogramTransformation
  • StarXterminator
  • Curves on the StarMask to increase the colours and that was it.
  • Curves on the background.
  • SCNR to reduce the green cast
  • UnsharpMask to increase detail
  • NoiseXterminator on the background only
  • PixelMath to recombine the stars with the background.

Here's what came out:

M106-session_1_session_2_session_3.thumb.png.af493ae08641caf0829f534e3b057083.png

Edited by Budgie1
Updated final image
  • Like 1
  • Thanks 1
Link to comment
Share on other sites

Oh wow - thank you all so much! Those images look amazing. What attracted me to M106 in the first place is the star colours along with the galaxy and those are captured really well :)

Ref flats, a master was included in the stack but I must confess I don't do flats very often and its probably time I generated some new ones, especially as I have now removed my CLS clip in LP filter since doing this image. That dust bunny is new, so something has shifted in my imaging train.

Last night I had a good play with DBE and did get better results that I was before, but still not perfect. Having seen one of Adam Blocks videos that suggested going more aggressive on the parameters as long as you ensure no stars are in your sample points (which I was doing already). Running the Canon banding script also seemed to help. Next step I guess is some kind of noise reduction but I'm going to follow the steps you guys have done and see what I get.

Thanks
Ed

Link to comment
Share on other sites

Hi

Nice data. There is some good detail emerging in the big galaxy 

Our PI trial expired some while ago but we came to the conclusion that it deals well with data which is perfect or near perfect, but needs hours in front of a computer otherwise. Disclaimer: that is almost certainly down to our lack of both skill and patience!

A few pointers:
- go through the frames and remove any which have cloud
- apply matched flat frames

I'll leave it to the PI gurus to advise exactly how to tackle stuff like this. We never found a way. Maybe do the heavy lifting elsewhere and come back to PI for the finishing touches? Dunno.
Anyway, 10 minutes in a different app but with no finishing touches gave...

Cheers and HTH
M106-session_1_session_2_session_3a.thumb.jpg.a2cecf55c9ed3be776956dc8de28c5b0.jpg

  • Thanks 1
Link to comment
Share on other sites

2 hours ago, edarter said:

Ref flats, a master was included in the stack but I must confess I don't do flats very often and its probably time I generated some new ones, especially as I have now removed my CLS clip in LP filter since doing this image. That dust bunny is new, so something has shifted in my imaging train.

Flats can be reused for different sessions, I do it myself, but only as long as nothing has changed or moved in the image train and the old Flats work. Changing a filter, removing the camera, changing it's orientation or (sometimes) just refocusing can all effect how old flats work during calibration.

If that's a new dust bunny then maybe time for a clean & new Flats. :D

  • Like 1
Link to comment
Share on other sites

  • 2 weeks later...

Thank you to everyone for your help on this one. Still a big learning curve for me but I'm getting there. Attached is my finished image. Not 100% happy with the background, still gradients in there which I've disguised a bit by bringing down the black point. Hopefully not too excessively though! My processing steps were:
 

  • Dynamic Crop
  • DBE x2
  • Background Neutralisation
  • Photometric colour calibration
  • Canon banding removal
  • TGV Denoise
  • Multiscale Median Transformation (for medium/low scale NR)
  • HistogramTransformation
  • StarNet2
  • SCNR on galaxy image to remove green cast
  • Curves with masks on the galaxy image to tame background and enhance galaxy / boost colour
  • UnsharpMask to increase detail
  • Curves on the star mask to boost colours
  • Morphological transformation on star mask to reduce star sizes
  • PixelMath to recombine the stars with the star mask
  • ADCNR as final touch of NR
  • Curves to tame the background further and lightly boost galaxy contrast

Now to decide the next target!

Thanks
Ed

M106 finished.jpg

  • Like 3
Link to comment
Share on other sites

@edarter Nice. That’s worked. Good to see the finished result when someone is helped on this site. I like the way you’ve highlighted star colour in the brighter stars but not the fainter ones.  There’s some nice detail too in the galaxies when you zoom in. 

Did you find morphological transformation worked for you? I’ve never got anywhere with it. It either makes my stars look horrible or when I reduce the effect it does nothing for the stars whatsoever.

I’ve taken to using Bill’s star reduction method which is a free add on to PI. Install and how to use explained In this YouTube video.  You need StarNet++ installed though. Also free. 

Link to comment
Share on other sites

7 hours ago, Ouroboros said:

@edarter Nice. That’s worked. Good to see the finished result when someone is helped on this site. I like the way you’ve highlighted star colour in the brighter stars but not the fainter ones.  There’s some nice detail too in the galaxies when you zoom in. 

Did you find morphological transformation worked for you? I’ve never got anywhere with it. It either makes my stars look horrible or when I reduce the effect it does nothing for the stars whatsoever.

I’ve taken to using Bill’s star reduction method which is a free add on to PI. Install and how to use explained In this YouTube video.  You need StarNet++ installed though. Also free. 

Thank you, I'm pretty happy with it but I suspect I will come back to it at some point and have another go at getting that background right 🙂

Ref MT, I wasn't too sure about it tbh. Seemed to reduce the small stars quite a bit but no impact on the big ones. I must confess I moved on from this part of the process quite quickly though as I was getting impatient for the end result. I do want to give Bill's method a go though. 

Thanks

Ed

  • Like 1
Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...

Important Information

We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue. By using this site, you agree to our Terms of Use.