DPChallenge: A Digital Photography Contest You are not logged in. (log in or register
 

DPChallenge Forums >> Individual Photograph Discussion >> Custom white balance in postprocessing?
Pages:  
Showing posts 1 - 11 of 11, (reverse)
AuthorThread
11/18/2005 05:41:33 PM · #1
This is a mainly hypothetical question but i have an image to go with it. Here's an image taken without custom white balance set:

I then shot a piece of white paper in the same light, gaussian blurred it into oblivion, and the average colour is R:145, G:90, B:43. After using this frame as a custom white balance the following shots came out correctly balanced.

So: any recommendations for how i would (mathematically rather than by trial and error) adjust the colour balance of the image in reference to the known white slide colour (above)?
11/18/2005 07:35:48 PM · #2
Note sure if this is what you are looking for but.
This was done by using levels.


YOu can See a guide of this here:
Guide

This was done by only making slight adjustments to the red curve:


Not sure if either of these help.

11/18/2005 08:35:57 PM · #3
agdcolor
This is a ps plug-in from which you might obtain a numbered reference to color temperature settings. A color balance plug is also on this site. Not exactly what I think your looking for, but related. I have not heard of a universal colour balance calculator or method - except for maybe a saved cb setting which you might use for a particular series of pictures made in the same conditions - with which you have to initially make the settings by eye. I suppose there might be some advantage to a mathematical system, if there is one, it sounds complex.

Message edited by author 2005-11-18 21:13:51.
11/18/2005 09:06:39 PM · #4
What about just shooting raw?
11/19/2005 08:17:23 AM · #5
Thanks for all the helpful responses, but none of you seem to have really understood exactky what i meant - i'm looking to be able to do this in postprocessing only, not calculate colour temperature in order to set white balance, and not do do it in levels by trial and error, but to work out a system (possibly using the channel mixer) to do this by a mathematical rule.

Maybe there's another way to put it - can anyone suggest a way of using the channel mixer or equivalent to set a specific colour temperature? It's basically the same thing i'm asking, except i'd have to calculate colour temperature in kelvin (which i can't measure directly) from the average white frame RGB value (which i can measure precisely).

As i say, it's a mainly theoretical point, so setting custom white balance in camera is a much more practical way of doing it.

Edit: that plugin would probably do just what i want, but a) i don't use photoshop and b) i'm trying to figure out a way to do this by hand

Message edited by author 2005-11-19 08:19:20.
11/19/2005 10:44:36 AM · #6
I would have to agree with undieyatch with out having any proof or formulas that it would be tough if not impossible to have one numerical balance to adjust a picture. since every shot would be captured in a different style of lighting with different intensities.
Then adding the variable of how the Camera interpreted the Lighting and the Formula it uses to balance the lighting, I would think this would vary greatly from camera to camera and shot to shot.

I too do not use Photoshop, But im sure you have heard the small number of the less fortunate ones in here mention Gimp.
THis program is free and does a lot of things photoshop does with some variations. It can also use plug ins.
From here I'll let this thread Die a less we have a Mathematical photographic geniuss that has the Real answer with data to back it.


11/19/2005 10:48:24 AM · #7
I don't know if this wil help, but have you tried shooting in RAW as one of the other posters mentioned. If, for example, you use Capture One, you can do an auto white balance on your 'white' frame and then apply exactly the same corrections to as many other frames as you wish. It also has kelvin colour temperature adjustment. I imagine that you could do pretty much the same in RAWShooter Essentials, which is priced a little more competitively.

Doing it in RAW therefore means that you don't have to set a custom white balance in camera, rather it can be set later and can be set as accurately as you like, so long as you shoot your reference frame.
11/19/2005 05:56:21 PM · #8
Originally posted by AJAger:


Doing it in RAW therefore means that you don't have to set a custom white balance in camera, rather it can be set later and can be set as accurately as you like, so long as you shoot your reference frame.


Yes yes yes, i understand and do all this, but as i say this is a mainly hypothetical question. I am certain there is a way to pull this off... i've tried using the channel mixer to adjust the red, green and blue to percentage amounts based on 100% being the highest number in the rgb value of the average white frame, which is a pretty simplistic way to normalise it but does the job. Of course it involves dropping two colours out of three somewhat, so the resulting image looks quite dark.

I'd just be fairly interested in the actual mathematics a camera does when it sets a custom white balance, or indeed the algorithm used to make a raw image conform to a certain colour temperature.
11/19/2005 06:08:42 PM · #9
This is my best attempt at a mathematical approach;



Put these % into channel mixer for each of the R, G and B channels and it should balance (haven't tested it, but that's the theory!)

Problem is that the Blue % is off the scale (max % is 200 in channel mixer) - So there's a serious lack of blue in that image to start with.

11/19/2005 06:37:22 PM · #10
Well, that's a totally new way to do it... the percentages are a bit funny, because of the 200% at least, but deviation from average is a cunning idea.

The maths i'd done previously was:
R 145/2.12 = 52%
G 165/2.12 = 77%
B 212/2.12 = 100%
Which yields:

The white balance looks about right to me, which leads me to think this is the right way of thinking, but obviously it's far too dark and adjusting brightness or intensity just brings back ghosts of the odd colour tinge.

Doing it your way however... well, the channel mixer isn't percentage precise anyway, but normalising it to fit 200% at most we get adjustment values of R:59, G:95, B:200 which comes out as:

It's a bit better but not perfect. What gave you the idea of using the
mean though?
11/19/2005 06:39:09 PM · #11
Originally posted by riot:

What gave you the idea of using the mean though?

Because I tried your method first, and it was too dark! :) So I thought find the average, and then adjust up/down from there.
Pages:  
Current Server Time: 04/19/2024 06:34:17 AM

Please log in or register to post to the forums.


Home - Challenges - Community - League - Photos - Cameras - Lenses - Learn - Prints! - Help - Terms of Use - Privacy - Top ^
DPChallenge, and website content and design, Copyright © 2001-2024 Challenging Technologies, LLC.
All digital photo copyrights belong to the photographers and may not be used without permission.
Current Server Time: 04/19/2024 06:34:17 AM EDT.