#1  
Old 19-06-2021, 02:21 PM
vlazg's Avatar
vlazg (George)
Registered User

vlazg is offline
 
Join Date: May 2014
Location: Darwin
Posts: 737
Synthetic luminance ?

Could someone tel me the pros and cons of obtains a synthetic luminance from RGB, and if so, do you align and then stack all subs?
Is a proper luminance filter better?

George
Reply With Quote
  #2  
Old 25-06-2021, 10:25 PM
Ryderscope's Avatar
Ryderscope (Rodney)
Registered User

Ryderscope is offline
 
Join Date: May 2006
Location: Glanmire, NSW
Posts: 2,169
Hi George,
There are three questions raised here so I shall attempt to respond to each. My position is that I don’t take Luminance any more, I create a synthetic luminance from the RGB data. I then process this as I would a ‘real’ luminance with the objective of bringing out structure and fine detail prior to applying an LRGB combination.

Do you align and stack all subs?
If you choose to use a synthetic luminance there are a number of ways that can be deployed. Some will do as you said, align and stack all subs. I don’t do that, I simply take the integrated masters for each of R,G & B and integrate them without any pixel rejection applied. I use Pixinsight for my processing software and have used the Sub Frame Selector process to measure the noise levels and SNR weights of synthetic luminance against captured luminance and it is usually better.

Is a luminance filter better?
Technically the answer is maybe but it depends on the band pass of the luminance filter. If you look at the spectrum graph for a set of LRGB filters typically you will see that the luminance filter covers a slightly wider band than the three RGB filters. You could argue therefore that the luminance data contains more information than the combined RGB filter set. So a synthetic luminance then would be lacking. I have not done any objective tests to try and quantify any losses but from a subjective image quality assessment there does not seem to be any disadvantage. I can report though a definite improvement in SNR with the application of a synthetic luminance.

Pros and Cons of Synthetic Luminance?
Traditionally we captured lots of luminance data and only sufficient RGB data to add chrominance to the image. Our eyes perceive structure and detail through the luminance so this is important. Typically as well, the RGB was captured as binned (x2) data which I presume stemmed from saving imaging time. This would be particularly so if one has to travel to a dark sky location for imaging. This being the case, one of the benefits of capturing data in this way is saving imaging time particularly when dark sky access is precious.

The alternative is to capture RGB unbinned and extract a synthetic luminance from the RGB. In my case I’m lucky in that I have my own observatory and don’t have to travel to collect my data.

I am not convinced that the supporting argument of saving imaging time by capturing luminance and binned RGB holds. My view is that if you apply the equivalent imaging time to capturing unbinned colour data you can produce relatively noise free colour channels and a synthetic luminance that has a high SNR. FWIW, my objective for most targets with my QSI683/Tak TSA 120 combination is to grab 30 subs of 5 minutes for each of RGB.

Of course, I don’t pretend to have the definitive answer to this and there are others that will no doubt want to contribute more.

CS,
Rodney
Reply With Quote
  #3  
Old 26-06-2021, 07:26 AM
vlazg's Avatar
vlazg (George)
Registered User

vlazg is offline
 
Join Date: May 2014
Location: Darwin
Posts: 737
Thanks Rodney for your answer.. imaging time is not a concern this time of year up here, the nights are monotonously clear for a few months ( it rained last night…) and my site is 95 k out of Darwin ,an hours drive , no big deal..
If the quality of lum matches the quality of synthetic lum then ok I’ll try it.

I have read that the unweighted average of individual RGB components does not give an optimal snr, different weights for each colour based on scaled noise estimates should be assigned.
Pix Image integration tool
Select RGB files
Parameters default
Combination-average
Normalisation = additive with scaling
Weights=noise evaluation
Scale estimator=iterative k sigma
Generate integrated image = enabled
Evaluate noise = enabled
Pixel rejection = no rejection
Clip low range =disabled
Clip high range = disabled

This came from Jim’s Pixinsight crib sheet.

Thanks again Rodney.

Any others ?

George

Last edited by vlazg; 26-06-2021 at 07:39 AM.
Reply With Quote
  #4  
Old 26-06-2021, 09:18 AM
gregbradley's Avatar
gregbradley
Registered User

gregbradley is offline
 
Join Date: Feb 2006
Location: Sydney
Posts: 17,877
I haven't found much benefit in synthetic luminance from RGB filtered images.

I think its better to simply take more luminance.

I usually bin my RGB so it also tends to be lower resolution.

Also colour is often taken when the object has moved away from the zenith as you concentrate on getting the sharpest luminance images first unless you do luminance on one or several nights and colour on another.

There's no substitute for high QE cameras, fast optics with large aperture and good seeing with dark skies plus good tracking.

That technique was developed at a time when the typical CCD had 50% QE or less and optics tended to be slow. Its a bandaid solution.

The best images have the best basics in them that were done very well not processing trickery which always shows through like cosmetic surgery always seems to.

Greg.
Reply With Quote
  #5  
Old 26-06-2021, 01:23 PM
Ryderscope's Avatar
Ryderscope (Rodney)
Registered User

Ryderscope is offline
 
Join Date: May 2006
Location: Glanmire, NSW
Posts: 2,169
Quote:
Originally Posted by vlazg View Post
Thanks Rodney for your answer.. imaging time is not a concern this time of year up here, the nights are monotonously clear for a few months ( it rained last night…) and my site is 95 k out of Darwin ,an hours drive , no big deal..
If the quality of lum matches the quality of synthetic lum then ok I’ll try it.

I have read that the unweighted average of individual RGB components does not give an optimal snr, different weights for each colour based on scaled noise estimates should be assigned.
Pix Image integration tool
Select RGB files
Parameters default
Combination-average
Normalisation = additive with scaling
Weights=noise evaluation
Scale estimator=iterative k sigma
Generate integrated image = enabled
Evaluate noise = enabled
Pixel rejection = no rejection
Clip low range =disabled
Clip high range = disabled

This came from Jim’s Pixinsight crib sheet.

Thanks again Rodney.

Any others ?

George
The settings that you list there George are the exact same ones that I use for creating a synthetic luminance.

I'm pleased that you are getting some clear skies up north as I cannot say the same for us here in NSW.
Reply With Quote
  #6  
Old 26-06-2021, 02:38 PM
Ryderscope's Avatar
Ryderscope (Rodney)
Registered User

Ryderscope is offline
 
Join Date: May 2006
Location: Glanmire, NSW
Posts: 2,169
Quote:
Originally Posted by gregbradley View Post
I haven't found much benefit in synthetic luminance from RGB filtered images.

I think its better to simply take more luminance.

I usually bin my RGB so it also tends to be lower resolution.

Also colour is often taken when the object has moved away from the zenith as you concentrate on getting the sharpest luminance images first unless you do luminance on one or several nights and colour on another.

There's no substitute for high QE cameras, fast optics with large aperture and good seeing with dark skies plus good tracking.

That technique was developed at a time when the typical CCD had 50% QE or less and optics tended to be slow. Its a bandaid solution.

The best images have the best basics in them that were done very well not processing trickery which always shows through like cosmetic surgery always seems to.

Greg.
I admit to still having an open mind on this Greg so I'm looking to test my arguments in support of synthetic luminance. I may just learn something in the process This being the case, let me unpack your response if I may.

I haven't found much benefit in synthetic luminance from RGB filtered images.
I think its better to simply take more luminance.
I usually bin my RGB so it also tends to be lower resolution.


This is the core issue as to whether there is a quantitative benefit to be realised from the two broad options:
1. Unbinned colour with synthetic luminance; or
2. Binned colour and unbinned luminance.

I see that you advise little benefit from synthetic luminance but I would need to understand what this means. Was the luminance extracted from binned data etc? How was the result evaluated? etc.

... colour is often taken when the object has moved away from the zenith ... concentrate on getting the sharpest luminance images first ...

A good approach however I would argue that imaging in RGB whilst spreading the acquisition evenly throughout the imaging times would ensure that we have data that is subject to the same conditions.

There's no substitute for high QE cameras, fast optics with large aperture and good seeing with dark skies plus good tracking.

Well I certainly cannot argue with that logic. However, I cannot see how that point is relevant. If we are looking for a quantitative assessment of two options then we must keep the other variables consistent. Therefore, for the same equipment configuration, sky conditions, tracking etc., are we better off with RGB unbinned and a synthetic luminance?

That technique was developed at a time when the typical CCD had 50% QE or less and optics tended to be slow. Its a bandaid solution.

I assume that the words "that technique" here refer to the application of a synthetic luminance. I must admit that I haven't seen any reference to this being developed as a 'band aid solution' necessary to overcome limitations with optics and camera sensors. In fact, I would have thought that this is a technique developed over the past decade to take advantage of the improvements in camera filter technology. I would be interested in any reference material you have on this.


The best images have the best basics in them that were done very well not processing trickery which always shows through like cosmetic surgery always seems to.

I don't agree that terminology such as "processing trickery" and "cosmetic surgery" are applicable here in the context of synthetic luminance as they don't really add any value in terms of why one method would be better than the other. Either there is something that we can see and/or measure or there isn't.

I'm looking forward to a continuing discussion on the pros and cons that properly evaluates the different options. To this end I will see if I can hunt down some data where I have captured luminance as well as unbinned RGB and post them here for evaluation.

CS,
Rodney
Reply With Quote
  #7  
Old 26-06-2021, 04:39 PM
vlazg's Avatar
vlazg (George)
Registered User

vlazg is offline
 
Join Date: May 2014
Location: Darwin
Posts: 737
Quote:
Originally Posted by Ryderscope View Post

I'm pleased that you are getting some clear skies up north as I cannot say the same for us here in NSW.
Sorry about the skies down there Rodney, I know how you feel having lived in Bathurst for 2years and then Lithgow for30. I do not miss the winters though
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 10:16 AM.

Powered by vBulletin Version 3.8.7 | Copyright ©2000 - 2024, Jelsoft Enterprises Ltd.
Advertisement
Testar
Advertisement
Bintel
Advertisement