Linear fit luminance and other linear fit questions ???

stevek

Well-known member
Jun 21, 2013
157
1
Hi,
I normally linear fit the R,G,B channels to the lowest median value channel of those channels (as shown in statistics). By linear fitting like this I find colour calibration is normally not necessary, even though I still carry it out.

Luminance is normally much higher R,G,B. Should the luminance be similarly linear fitted to the RGB channels as well? Or would one be "wasting" much of the integration time if doing that? I have started to do a linear fit of the luminance but I am not 100% sure about this and whether it is valid.

I normally do NOT linear fit the Ha to the LRGB channels. I'd only even linear fit the Ha to the OIII if doing a HOO pallette.

Can anyone add any commentary to what I am doing please?
 

Juan Conejero

PTeam Member
Sep 2, 2004
7,405
68
56
Valencia, Spain
pixinsight.com
That does not make any sense. Individual RGB channels and luminance components, as well as any other image components, should not be mutually fitted. Or at least, I can't see any valid reason, under normal working conditions, to perform such fittings.

Why are you doing this? Where does the idea of mutually fitting channels and luminance components come from?
 

pfile

PTeam Member
Nov 23, 2009
5,324
64
there are some tutorials out there that talk about doing this. i don't know why. in other forums i try to explain that any color balancing step that follows the linear fit is just going to undo the linear fit anyway, so why bother?

just about the only reason i can think of to do something like this on an RGB image is to avoid having one super-bright channel while doing DBE. that bright channel would force you to set the sample tolerance really high and then perhaps oversample the dimmer channels. but this would only apply to an image from an OSC camera. if you have a mono camera then you should do DBE on each channel independently.
 

ngc1535

PTeam Member
Feb 1, 2014
506
54
FINALLY this is being discussed. I have a "screed" in my tutorials:


There are many tutorials, almost virally so, that say to mutually fit color channels as a matter of procedure. This topic was also one of the earliest questions I ever submitted to this forum. Needless to say, I have a thing about this. :)


In my post Juan addressed how LinearFit works- but did not address how it would be useful to do this to "mutually fit" them which was what I was trying to ask back then.

-adam
 

stevek

Well-known member
Jun 21, 2013
157
1
Hi, thanks Adam. I actually subscribe to your fundamentals section but, for right or wrong, I did not watch this section thinking (erroneously as it appears) that I "knew it all" when it comes to linear fit. Clearly I don't and so I will watch those materials.

As Rob says, I find I can sometimes have issues with DBE unless I do linear fit beforehand. It also seems to semantically make sense to balance the channels right off the bat.......
-steve
 

billx

Member
Aug 3, 2019
8
0
Please explain this process more. I am a relatively new user, and have seen tutorials that do the above. I recently had some data and I couldn't remove some gradients easily, and when I used Linear fit as above, it solved the problem. I was using OSC and separated the channels to do Linear fit. Please enlighten me as to why this is not a good technique.
 

pfile

PTeam Member
Nov 23, 2009
5,324
64
? linear fit itself wont remove gradients but as i said above on an RGB image it might be useful prior to DBE/ABE. alternatively of course you can always split your image into its 3 channel components and run DBE on each one separately and then reassemble the image. that is no different than what happens when someone with a mono camera does DBE prior to channel combination.
 

billx

Member
Aug 3, 2019
8
0
Thank you - I did not think of using DBE on each channel and then recombine. It's just that when I did the Linear fit on each channel, and then recombined them, and then did DBE on the recombination, the image looked better than if I just ran DBE on the regular combined image. I just figured it had something to do with the Linear fit process. I will now try using DBE on each channel first.