Jump to content

  •  

CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.

Photo

Normalize Scale Gradient Alternative

  • Please log in to reply
7 replies to this topic

#1 FiveByEagle

FiveByEagle

    Messenger

  • *****
  • topic starter
  • Posts: 456
  • Joined: 04 Mar 2020

Posted 22 June 2021 - 11:08 AM

Hello all!

 

Backstory: I stack with APP or DSS and then import into Pi for all processing, and then dump into Photoshop for finalizing and distribution. 

 

Story: Pixinsight just added the NSG (I'm using that from now on) script recently, and in some testing on my newest datasets it absolutely works wonders. Re-stacked the data and they required no LPR in APP or DBE in Pi - they were clean and ready for processing. 

 

Fact: I drizzle my data during integration. New-ish OTA's pixel scale is 1.02 and many of my test images had blocky stars, so one day I turned on Drizzle in DSS/APP and that issue went away and the image was noticeably better. In order to test if I was fooling myself, I walked away for 24 hours and came back to the data and it was apparent side-by-side the drizzled stack was better.

 

Issue: NSG does not support Drizzle, and from the Pi forum posts it'll probably be a while before it is considered. 

 

Question: Is there any normalization technique in APP or DSS that does a similar thing to NSG only that I can still Drizzle 2x? I have tried looking around prior to posting but there does not seem to be any clear technique.

 

 

 



#2 R Botero

R Botero

    Gemini

  • -----
  • Posts: 3,355
  • Joined: 02 Jan 2009
  • Loc: Kent, England

Posted 23 June 2021 - 05:44 AM

Short answer is NO.

 

Long answer is you could still try to generate drizzle data (xdrz files) from your Image Alignment process, run the aligned frames through NSG, integrate manually to update the xdrz files and then use those for Drizzle Integration.  See what comes out.

 

Roberto



#3 johnpane

johnpane

    Mariner 2

  • *****
  • Posts: 278
  • Joined: 13 Dec 2014
  • Loc: Wexford, PA, USA

Posted 23 June 2021 - 07:10 AM

Long answer is you could still try to generate drizzle data (xdrz files) from your Image Alignment process, run the aligned frames through NSG, integrate manually to update the xdrz files and then use those for Drizzle Integration.  See what comes out.

This will not result in DrizzleIntegration using the normalized subframes. Those are deBayered but drizzle uses the original CFA files for image data.  A more complex approach as laid out by John Murphy (NormalizeScaleGradient author) might work but I am not aware that anyone has confirmed that yet. 



#4 FiveByEagle

FiveByEagle

    Messenger

  • *****
  • topic starter
  • Posts: 456
  • Joined: 04 Mar 2020

Posted 23 June 2021 - 08:04 AM

This will not result in DrizzleIntegration using the normalized subframes. Those are deBayered but drizzle uses the original CFA files for image data.  A more complex approach as laid out by John Murphy (NormalizeScaleGradient author) might work but I am not aware that anyone has confirmed that yet. 

Might give this a shot- When Pix gets into the weeds (kinda where we are rn) I get a little turned around.

 

I can't wait for this to be added. Half of my processing is removing difficult gradients, and my last few stacks with NSG just "didn't" have any. versus just using WBPP on its own.


  • limeyx likes this

#5 R Botero

R Botero

    Gemini

  • -----
  • Posts: 3,355
  • Joined: 02 Jan 2009
  • Loc: Kent, England

Posted 23 June 2021 - 10:53 AM

I was under the impression that only .xdrz files were necessary for Drizzle Integration - as per Juan Conejero's post https://pixinsight.c...3318/post-82576  with those files containing all the information about pixel values, alignment transformation and pixel rejection statistics.

 

Since the NSG script changes the pixel values in the .xisf files it processes, my assumption was that by integrating the corrected and properly weighted files, the updated .xdrz files would reflect the new values (and weight/pixel rejection factors).

 

I am confused by the need for .xnml files mentioned by John Murphy in the post referenced above since those are Local Normalization files rather than drizzle (.xdrz) files.   Maybe just a typo.

 

I have been using the NSG script since John first made it public with great results but I don't drizzle (anymore) as I am not particularly undersampled.  I may try what I outlined next time I can to see if the drizzled integration reflects the normalisation.

 

Roberto



#6 johnpane

johnpane

    Mariner 2

  • *****
  • Posts: 278
  • Joined: 13 Dec 2014
  • Loc: Wexford, PA, USA

Posted 23 June 2021 - 11:15 AM

I was under the impression that only .xdrz files were necessary for Drizzle Integration - as per Juan Conejero's post https://pixinsight.c...3318/post-82576  with those files containing all the information about pixel values, alignment transformation and pixel rejection statistics.

 

Since the NSG script changes the pixel values in the .xisf files it processes, my assumption was that by integrating the corrected and properly weighted files, the updated .xdrz files would reflect the new values (and weight/pixel rejection factors).

 

I am confused by the need for .xnml files mentioned by John Murphy in the post referenced above since those are Local Normalization files rather than drizzle (.xdrz) files.   Maybe just a typo.

 

This is answered in post 3 above. NSG does not alter the original CFA files that are used by DrizzleIntegration. The normalized subframes have already been deBayered and thus cannot be used by DrizzleIntegration. The .xnml files would be a way to convey the normalization parameters to DrizzleIntegration so it can apply them to the CFA data.



#7 R Botero

R Botero

    Gemini

  • -----
  • Posts: 3,355
  • Joined: 02 Jan 2009
  • Loc: Kent, England

Posted 23 June 2021 - 11:39 AM

John,

 

The simple answer is here:  https://pixinsight.c...507/post-101799  I posted my argument to the PixInsight forum where NSG is being discussed and Rob clarified the issue for me. I don't think it matters if the underlying frames are colour and need CFA (colour) treatment or not.

To the OP, try the hack posted by John and let us know.

Thanks

 

Roberto



#8 johnpane

johnpane

    Mariner 2

  • *****
  • Posts: 278
  • Joined: 13 Dec 2014
  • Loc: Wexford, PA, USA

Posted 23 June 2021 - 11:49 AM

Ah, yes, my thinking was centered around my own workflow with OSC images, where the CFA issue does matter. But you're right that particular issue is not present in other workflows.




CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.


Recent Topics






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics