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

Stacking all my lights together or as a stack of substacks

  • Please log in to reply
4 replies to this topic

#1 smiller

smiller

    Gemini

  • *****
  • topic starter
  • Posts: 3,123
  • Joined: 27 Oct 2018
  • Loc: Vancouver Washington (not BC!)

Posted 03 April 2024 - 09:47 AM

This is probably only of interest to a few, but I thought I would share a process I'm using to deal with a zillion lights.

 

With my short exposures resulting in thousands of lights and my Alt/Az field rotation, there are some real advantages in SIRIL in stacking my lights in 10 or 20 subsets and then stacking those results together.  Spot testing indicates the quality is, on average, equivalent.  Below a small test with the Hidden Galaxy showing stacking 20 subsets (20 sets of 72 * 8 Second Lights) and then stacking those together versus stacking all lights at once.  

 

I've done about 4 tests and found that they are materially equivalent.  There are always very minor differences in noise and details but once finally processed with equivalent levels of processing to the best of my ability, they are generally indistinguishable.  Any remaining differences are much less than minor editing preferences like small adjustments of sharpening applied or the amount of noise reduction... so even minor editing difference tweaks are a much greater impact.  In other words, the remaining differences are in the noise, so to speak.  So this indicates to me for all practical purposes, stacking as substacks and stacking them all together is equivalent enough to be a good option for me.

 

Here is one example of the core of the Hidden Galaxy taking a couple of nights ago and just roughly processed:

 

You may have to click on the link to get the .GIF to animate:

Also, I did notice PIPP compressed the GIF just a tad and there is a tiny bit of posterization (quantization of light levels) in the bright core.  That is not a stacking artifact.

 

get.jpg?insecure

 

 

A couple more details on stacking the substacks:

1) I crated the substacks with SIRIL.

2) I stacked the substacks with both SIRIL and APP to compare.  They were equivalent.  I used weighting based on quality in APP and basded on SNR in SIRIL.

3) I did light processing on each with the same parameters: BXT, Stack based color correction, Some NXT, and finally a tweak of stretch and color correction to make them roughly equal in brightness, contrast, and color cast to make comparison easier.

 

For me there are a couple implications:

 

1) I can do this on my long captures with thousands of lights to reduce stacking time and disk space requirements.  It turn out that these aren't completely linear, especially for Alt/Az captures with field rotation over time, so stacking from substacks can be several times faster and 1/2 the disk space.

2) I can save these substacks as "virtual long lights" of ,say, 10 or 20 mintues each, to then more easily integrate together with other sub-stacks on multilight captures as I add more data later or in the following years. 

 

Finally, I have SIRIL scripts to do this all other than splitting th lights up.  I just place 5% or 10% of the lights in each of the 10 or 20 predefined directories in a predefined folder (light1, light2, light3...) along with the Biases and Flats and run the script.   SIRIL creates one set of Biase and Flats masters and then stacks each directory and then stacks all these results together.  So it only takes 2-3 minutes to setup to run.

 

- Steven


Edited by smiller, 03 April 2024 - 11:03 AM.


#2 Charlie B

Charlie B

    Surveyor 1

  • -----
  • Posts: 1,703
  • Joined: 22 Mar 2008
  • Loc: Sterling, Virginia

Posted 03 April 2024 - 01:46 PM

Have you tried fast imaging stacking in Pixinsight?  It would be nice to compare.

 

Regards,

 

Charlie B



#3 smiller

smiller

    Gemini

  • *****
  • topic starter
  • Posts: 3,123
  • Joined: 27 Oct 2018
  • Loc: Vancouver Washington (not BC!)

Posted 03 April 2024 - 04:38 PM

Have you tried fast imaging stacking in Pixinsight?  It would be nice to compare.

 

Regards,

 

Charlie B

I have, when stacking a lot of lights, I round results to be roughly similar to SIRIL, but SIRIL is still much faster because SIRILs calibration and registration are so much faster, which is still slow in PI.   Still the Fast Stacking feature in PI is a great advancement for those with a lot of subs.



#4 AaronH

AaronH

    Viking 1

  • *****
  • Posts: 873
  • Joined: 20 Apr 2021
  • Loc: NSW, Australia

Posted 03 April 2024 - 07:12 PM

This "virtual long lights" approach seems sensible, and it's hard to think of any real disadvantage.

 

The only case where it may be a problem is if your data is very inconsistent frame-to-frame. Having a larger set to weight from will allow you to penalise and reject outliers more easily. However, if there are no massive frame-to-frame variations, or if you have an effective way to detect and reject outliers within your subsets, then this seems like a great approach for dealing with massive numbers of frames.


  • smiller likes this

#5 smiller

smiller

    Gemini

  • *****
  • topic starter
  • Posts: 3,123
  • Joined: 27 Oct 2018
  • Loc: Vancouver Washington (not BC!)

Posted 03 April 2024 - 07:18 PM

This "virtual long lights" approach seems sensible, and it's hard to think of any real disadvantage.

 

The only case where it may be a problem is if your data is very inconsistent frame-to-frame. Having a larger set to weight from will allow you to penalise and reject outliers more easily. However, if there are no massive frame-to-frame variations, or if you have an effective way to detect and reject outliers within your subsets, then this seems like a great approach for dealing with massive numbers of frames.

From the entire set, I do toss bad frames based on FWHM and starcount, so I don't have any stinkers in there.   Then they are stacked with weighted based on SNR.

 

I have considered also stacking the best 50% based on FWHM for achiving purposes if I ever wanted to revisit only stacking a high detailed set if I get another night that has better seeing.  




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