Custom D65 White Balance

[Juha_Lintula]

I have a D65 calibrated screen, and I took a picture of a white screen. I used the picture to set the initial white balance. I used the custom WB on an image of a color checker, and got a better delta E in color calibration. Now I’m wondering why WB module shows, with my custom coefficients temperature 5064K, and the standard coefficients show 6502K. This may be a stupid question but why dt shows color temperature of 5000K of an image of 6500K screen?

[kanyck (Nick)]

There was a discussion of this matter on the following thread:

Confused about D50, D65, and CCT in white balance and color calibration modules

I recently started using the color calibration module instead of the white balance module for all my images. One thing that really confused me was why the color temperature of everything seemed wrong. For example, a photo in direct noon sunlight looks balanced at 3972K5000K instead of 6500K, like I would expect. I finally realized that if I set the illuminant for that same photo to “same as pipeline (D50)” and back to “D (daylight)”, the temp will be 5002K and the color looks right. Seeing that …

[flannelhead (Sakari)]

Darktable’s pipe uses D50 as the reference white. This means that in the conversions between RGB and XYZ representation, RGB = (1, 1, 1) maps to D50. I believe this is the reason.

[Juha_Lintula]

I’m still confused why the default coefficients show temperature 6502K, and my shot of D65 white is displayed in the WB module 5064K. Are the default coefficients wrong, or should I take another picture where my screen is calibrated at another temperature.

PS. If someone else has an OM-1 and wants to try my customized WB coefficients they are here
R: 2.184
G: 1.000
B: 1.761

[flannelhead (Sakari)]

Apologies, I was a bit too fast to read your original post…

Keep in mind that a LED backlit computer display, even if calibrated to D65, is very far from real daylight in terms of the spectrum. So that’s why the temperature and tint shown in the WB module would be practically bogus.

I don’t see very much benefit in using custom WB coefficients acquired this way.

[Donatzsky (Nis Donatzsky)]

@flannelhead

I don’t see very much benefit in using custom WB coefficients acquired this way.

While not perfect, certainly, it massively improved things with my Canon 5D mk II.

[Juha_Lintula]

I also got slightly better ∆E in color calibration when I used the WB values calculator from the screen image.

[priort (Todd Prior)]

How is the lighting for the shot… Perhaps you have some ambient light effects…for sure it won’t be 6500 from a shot of the screen but in general for some reason it does help it seems for many users to adjust the red and help with color when you do this exercise

[geni1105]

@flannelhead

Apologies, I was a bit too fast to read your original post…

Keep in mind that a LED backlit computer display, even if calibrated to D65, is very far from real daylight in terms of the spectrum. So that’s why the temperature and tint shown in the WB module would be practically bogus.

I don’t see very much benefit in using custom WB coefficients acquired this way.

This procedure is recommended for Olympus cameras in the darktable manual. Not sure if it is still relevant - at least it did not improve color rendering for mine.

[MStraeten (MartinSt)]

it’s recommended there because there were a lot of issue reports regarding whitebalance. In this case that custom measurement is better then nothing

Just because there‘s a description how to generate a custom set of coefficients it isn’t an indication that the default coefficients generated by adobe are not good enough

In general it makes sense if you see issues with the default white balance that indicates tat your camera doesn’t behave like the majority of same camera model.