#1  
Old 16-04-2020, 03:24 PM
JA
.....

JA is offline
 
Join Date: Oct 2016
Location: Melbourne, Australia
Posts: 1,697
Noise Comparison Stacking Images

Hello Everyone,

I hope everyone is happy and all's well. I've always been interested in wanting to visualize the difference between stacking more and more sub-exposures, with a view to improving the signal to noise ratio in images. In particular how much better do things get for each bit of extra effort.

Well there is plenty of information/theory out there (simplifying) to suggest that given "appropriate" sub-exposures that signal to noise ratio can be improved by as much as the square-root of 2 (an ~41% improvement) for each doubling of the number of sub-exposures.

I set about to try to see what that looks like, not that I tweaked things to the max to achieve sky-limited exposures, but just "reasonable exposures" which were not obviously black clipped on the camera histogram. Using a Modified Nikon D800 with Sigma Art 50mm f/1.4 lens on a fixed tripod/no tracking I imaged the Milky Way from my suburban Melbourne backyard. Unfortunately as this was conducted last year and I've filed/misplaced the original image details, but most likely they were 8 second exposures @f/2 at ISO800. The images were stacked in Sequator, with its Light Pollution processing applied but no specific noise reduction was applied. The final n=1,2,4,8,16 and 30 final images were somewhat adjusted to equalise brightness to facilitate comparison.

The full image is attached of the stack of 30 x 8s images of the Milky Way. There is also a comparison below this image of the Cat's Paw Nebula magnified (Yellow Rectangle) to show detail for n = 1, 2, 4, 8, 16 and 30 sub-exposures, i.e: each time doubling the number of sub-exposures. (Yes i know it should have been 32). There is clearly a benefit with each doubling in the number of sub-exposures and 30 seems a decent/realistic endpoint given the overall image size/FOV and for the conditions. I'd be interested to continue it past 30 images to see the point of no further benefit. I might do that one day. In any event it's important to recognize that some of the benefit seen will also be due to some mild "dithering", given the use of a fixed tripod without tracking.

Happy viewing. (Image TBA soon)EDIT: Now added

Best
JA
Click image for larger version

Name:	Noise Comparison Stacking -JA.jpg
Views:	53
Size:	206.7 KB
ID:	257863
Attached Thumbnails
Click for full-size image (Noise Comparison Stacking -JA - Copy.jpg)
206.7 KB30 views

Last edited by JA; 16-04-2020 at 03:56 PM. Reason: image now added
Reply With Quote
  #2  
Old 16-04-2020, 08:52 PM
Atmos's Avatar
Atmos (Colin)
Ultimate Noob

Atmos is offline
 
Join Date: Aug 2011
Location: Melbourne
Posts: 6,858
Not sure how difficult or easy there are going to be to see but the first one shows a 1,4,8,16,32,64,127 (didn't have 128).

The second shows an enlarged version of 16,32,64,127 so show that there is indeed benefit for more and more exposure.
Attached Thumbnails
Click for full-size image (First Test.jpg)
128.3 KB34 views
Click for full-size image (Second Test.jpg)
109.3 KB33 views
Reply With Quote
  #3  
Old 16-04-2020, 09:04 PM
peter_4059's Avatar
peter_4059 (Peter)
Big Scopes are Cool

peter_4059 is offline
 
Join Date: Jun 2007
Location: Brisbane
Posts: 4,270
Worth checking out these cn posts by Craig Stark ...
https://www.cloudynights.com/article...t-part-1-r1895

https://www.cloudynights.com/article...ne-pixel-r1902

https://www.cloudynights.com/article...r-camera-r1929
Reply With Quote
  #4  
Old 16-04-2020, 10:19 PM
JA
.....

JA is offline
 
Join Date: Oct 2016
Location: Melbourne, Australia
Posts: 1,697
Thanks Peter. He in fact piqued my interest in the issue in the first place following one of his lectures.

Best
JA
Reply With Quote
  #5  
Old 16-04-2020, 10:21 PM
JA
.....

JA is offline
 
Join Date: Oct 2016
Location: Melbourne, Australia
Posts: 1,697
Quote:
Originally Posted by Atmos View Post
Not sure how difficult or easy there are going to be to see but the first one shows a 1,4,8,16,32,64,127 (didn't have 128).

The second shows an enlarged version of 16,32,64,127 so show that there is indeed benefit for more and more exposure.
Thanks Colin, Yes clearly of benefit, even though more subtle in your higher quality images.

Best
JA
Reply With Quote
  #6  
Old 17-04-2020, 05:53 AM
The_bluester's Avatar
The_bluester (Paul)
Registered User

The_bluester is offline
 
Join Date: Feb 2011
Location: Kilmore, Australia
Posts: 2,666
With my ASI294, I have found 100 subs to be a good target to aim for in terms of smoothing out the read and shot noise, 100 allows the use of light drizzle processing without reintroducing an objectionable amount of noise, at least for my camera.
Reply With Quote
  #7  
Old 17-04-2020, 12:23 PM
Tulloch's Avatar
Tulloch (Andrew)
Registered User

Tulloch is offline
 
Join Date: Sep 2019
Location: Melbourne
Posts: 260
I did a similar thing comparing the noise from a Canon 700D DSLR vs a ZWO ASI224MC for planetary imaging. Details here...

https://www.cloudynights.com/topic/6...ct-test-three/

The DSLR seems to have noise reduction built into the output frames, so require less frames to reduce the noise. It also runs a lot slower at 20fps, so needing less frames is advantageous.

Andrew
Reply With Quote
  #8  
Old 04-05-2020, 08:46 AM
JA
.....

JA is offline
 
Join Date: Oct 2016
Location: Melbourne, Australia
Posts: 1,697
Thanks Paul and Andrew for your insights and Andrew for the link to your own detailed comparison.

Best
JA

Last edited by JA; 09-05-2020 at 08:08 AM.
Reply With Quote
Reply

Bookmarks

Thread Tools
Rate This Thread
Rate This Thread:

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT +10. The time is now 02:25 AM.

Powered by vBulletin Version 3.8.7 | Copyright ©2000 - 2020, Jelsoft Enterprises Ltd.
Advertisement
Limpet Controller
Advertisement
Testar
Advertisement
Bintel
Advertisement
NexDome Observatories
Advertisement
Astronomy and Electronics Centre
Advertisement