Back to forums

LRGB VS L,R,G,B: Processing Comparison

Processing LRGB Deep Sky
6 replies•489 views
Which one is better?
Multiple choice poll 10 votes
60% (6 votes)
40% (4 votes)
You must be logged in to vote in this poll.
Alien_Enthusiast avatar
Alien_Enthusiast avatar
Over the past week, I've been collecting feedback regarding my latest ARP 273 image.

Many interesting ideas were proposed, which inspired me to try a different processing algorithm.

Instead of stretching the image after combining the LRGB channels, I tried stretching each channel individually and then combining them, as well as adding photometric color calibration.

To me, it seems like the new approach was more successful compared to my last attempt.

What do you guys think? Any advice would be greatly appreciated!


Here is the image comparison:

Well Written Insightful Respectful Concise Engaging
Jeff Horn avatar
Jeff Horn avatar
The images are a bit small on my monitor for intense scrutiny would be nice to link to larger images.  Maybe I don't follow but are you using PCC in the second example on a nonlinear combination?  Not sure that PCC is designed to work on nonlinear data.  Maybe there has been some update I'm not aware of .... look almost at the end of the documenation.
https://pixinsight.com/tutorials/PCC/
Andre Vilhena avatar
Andre Vilhena avatar
Hi,

PCC is designed to work on linear images, as it needs to measure the light fluxes to compute the proper calibration factors. It does not mean it won't give a result but it also does not mean it is a good result. Also, LRGB is designed to work on non-linear images.
So, it seems that neither of those combinations is technically correct. Why don't you try to do the correct process to compare? It should be:
- combine R, G and B and apply PCC
- stretch the RGB calibrated imagem
- stretch the L
- apply LRGBCombination

I'm skipping steps like gradient extraction and noise reduction.

Cheers,
André
Helpful Insightful Respectful
Brian Puhl avatar
Brian Puhl avatar
Simply put, that isn't how it works.   PC and SPCC are intended to work on linear data.  Combine RGB, calibrate data, perform a LINKED stretch.    Add your luminance data after stretching.  

It's not a matter of A or B, it's just a matter of doing it correctly.
George Hatfield avatar
George Hatfield avatar
I agree with Brian.  I create starless images of both the RGB and luminance (often a super luminance).  I stretch both, usually using GHS with or without arcsinh stretching.  I try to get the mean close for both images.  Then, I combine to form the LRGB.  Finally, I add the stars.

George

P.S. Of course, I color calibrate just after removing gradients (usually with GraXpert AI), from the RGB.  Then I run BX, NX, and StarX.  Then I stretch the starless image, and then I combine a similarly processed L with the RGB.  Once I have an acceptable LRGB starless image, I add back the stretched stars.   If there is a big difference between the R, G, and B, I run linear fit before combining to the RGB.
Helpful
Stefan Pfleger avatar
Stefan Pfleger avatar
Ok first of all:
SPCC MUST be done linearly. 
please please please dont apply it on a stretch!!!

secondly, i always stretch the RGB and then add the lum to it. I never combine the L + RGB in linear stage.
Alien_Enthusiast avatar
Alien_Enthusiast avatar
I have followed your advice and learnt how to stretch RGB and L separately and then combine as you recommended. Here are all 3 versions of the image.