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.


Lots more processing time, but 2x drizzle integration+CFA--> 2x downsampled resulting in significantly lower noise image

  • Please log in to reply
2 replies to this topic

#1 DrGomer


    Mariner 2

  • *****
  • topic starter
  • Posts: 220
  • Joined: 07 Aug 2019

Posted 20 February 2020 - 12:05 PM

Playing around with processing to see if I can increase S/N and came across an interesting result. Not sure if this is know. Not sure also why this is happening and if there are other tricks like this to improve S/N.

Any way I look at it, the results are VERY noticeable.  This process probably(guessing?) doesn't work well with the 20-50 sub stacks with long integration systems, but does with the typical cmos high # stack b/c of how drizzle works.

This is from ~425  120 sec subs with SV80+FF+ASI533 over multiple days (target hopping) with some moon light to contend with. ~14 hrs total.


Processed in PI.

1) Blink to reject any subs with significant tracking error or clouds

2) calibrate with Flats-darkflat and Darks. 

3) Debayer with VNG

4) Register and generate drizzle data

5) Integrate with Generalized Extreme Studentized Deviate rejection (default settings), noise evaluate weightings, and generate drizzle data checked to update drizzle files

This produces the LEFT image (DBE only applied with auto stretch on)

6) Drizzle integrate with  files produced from #1-5 with default settings (Scale=2, drop=0.9, square, pixel rejection and weighting =on).

6a) Also  enable CFA drizzle

7) Downsample integrated image by factor of 2 (2x2 bin averaging) so that we are back to original ~3000x3000 pixels.

This produces the RIGHT image (DBE only applied with auto stretch on)


attached images are zoomed in autostreched crops to show the raw noise around NGC 3628. No filtering, color calibration, curves, color correction etc.

CN will probably compress image, so also linking dropbox XISF files (106MB each)
Normal processing: https://www.dropbox....n_DBE.xisf?dl=0
+Drizzle processing: https://www.dropbox....ample.xisf?dl=0

Attached Thumbnails

  • Integration comparison.JPG

  • bmhjr and Peregrinatum like this

#2 bmhjr



  • ****-
  • Posts: 1,369
  • Joined: 02 Oct 2015
  • Loc: Colorado

Posted 20 February 2020 - 12:14 PM

Looks good to me!  I do the same as you have described except I will use Deconvolution on the Drizzle image before resampling back to original size.  It helps tighten the stars a bit and bring out some fine detail.

Edited by bmhjr, 20 February 2020 - 12:15 PM.

#3 Peregrinatum


    Viking 1

  • *****
  • Posts: 897
  • Joined: 27 Dec 2018
  • Loc: South Central Valley, Ca

Posted 20 February 2020 - 02:43 PM

I agree... dithering adds more time and makes me want to throw a brick at SGP sometimes waiting for it, but in the end the drizzle integration always look much better to me than without...


Try playing around with the drop shrink factor and using different functions like Gaussian, or K > 2, takes longer but looks really good...


Here is some good info on DI




I think DI is a must for anyone with LP issues.

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