Hello,
I was wondering if there is an easy way in RT how to obtain the best spot for white balance. Sometimes I find a picture problematic and in order to find the best spot for WB I must spent just too much time on looking for it (which is an area that has similar or close all the RGB values)
In gimp/photoshop it gets easier because I can create a new, blank layer, fill it with middle gray, set its blending mode to ādifferenceā and use threshold to obtain it (see: https://patdavid.net/2011/06/getting-around-in-gimp-color-correction/).
Is that technique possible in RT? I know that 'retinexā has some masking capabilities, but it is complicated tool. Iām currently experimenting with it and wavelets ā but with no luck.
If it is not possible, would it be difficult to implement such a nice feature? It would be really helpful and save rt users a great deal of time.
Another idea would be to set a white balance according to skin tones. Alex from magic lantern project implemented it once in ufraw (White balance on skintones)
Also, iām crossing my fingers for @jdc and his WB ārobustā research and work.
Although I have Photoshop and Lightroom, I keep coming back to my favorite duo ā rt + gimp!
Thanks,
@kazah7 Good question and ideas. I have done some reading on wb myself because it is so fascinating. There is lots of research being done that has resulted in numerous theories and methods. Some references for you to check out if you havenāt already:
The only thing that is pretty much right is the matrix operation between the illuminant and the base color found on the spectrometer.
But in the majority of cases, the problem is that we do not know the iluminant - if we knew it, the problem would be almost solved, nor the spectral color of the subject (flower, face, sky, etc.).
We are faced with a problem where there are more unknowns than equations
The alternative solutions presented in eg Photoshop or other software like Ufraw (patch for skin), are only work around and do not solve the problem.
But they give the impression of doing something serious, omitting a parameter from the equation
I do not neglect these solutions but in most cases they are false.
Creating a target gray mask (or other hue) is a solution to target / reach a color, but is generaly obsolete for the white balance.
Indeed the gray that we will target on the image itself depends on the coefficients of the white balance that is applied to the red, blue and green channels.
So in French ācāest le chat qui se mord la queueā; itās the cat that is biting its tail.
It is the same if one uses the skin, whose variation of hue is close to 1 radian, which is enormous.
As I said the real solution is to take the picture, as would Alfred Hitchkock who always appeared in his films, putting a gray chart incorporated into the scene.