White Balance readout difference in Adobe Camera Raw vs Capture One

[nilo]

I finally take time to ask this:

A difference 800/2.0 in the WB between ACR and C1! Not that it matters so much, because the actual rendering does not reflect that tremendous difference at all, just why?

For the record: In this instance, reading was taken on the second gray patch from the left on the RAW image of a CC24.

regards

[madmanchan]

The white balance readout in terms of temperature & tint depends on the camera profile (more specifically, the translation between so-called "camera neutral" values and temperature/tint values). Different raw converters use different profiles. This results in different readouts, even if image appearance is the same.

The process of reporting white balance values is non-standardized.

[nilo]

Thank you for your reply!

I must say, that I do not fully grasp the explanation

[...] more specifically, the translation between so-called "camera neutral" values and temperature/tint values [...]

Could you tell me more and/or point me to some info?

Thanks in advance

regards

[Bart_van_der_Wolf]

Could you tell me more and/or point me to some info?

Hi Nino,

The reverse calculation of a color to its Correlated Color Temperature (CCT), is reasonably well understood for colors not too distant to the black body locus, and is usually modeled after Robertson's algorithm. However, to arrive at an XYZ color to begin with, requires some assumptions about the tri-chromatic color primaries of a camera. After all, we are trying to derive a full-spectrum color temperature from only 3 integrated, and partially overlapping, color bands. So it is not an absolute/accurate calculation.

Cheers,
Bart

[nilo]

[...] So it is not an absolute/accurate calculation.[...]

That's an euphemism ;-) considering a huge 800/2 difference between two top RAW converters. As if they describe two completely different things.

[Bart_van_der_Wolf]

That's an euphemism ;-) considering a huge 800/2 difference between two top RAW converters. As if they describe two completely different things.

Hi Nino,

Not quite. They describe the same thing, but because of using different assumptions of the actual color the resulting numbers differ. Yet that doesn't prevent them from Click white balancing correctly, because that now starts from the same assumptions (color coordinates in a defined trichromatic color space). The solution depends on the assumption of an XYZ color from the Raw image data, and the outcome depends on which color is actually used. Use the same assumption, you'll get the same output from the algorithm.

The main thing is that the color temperature and tint values should not be taken as absolute numbers, but rather as an approximation, and relative differences within that application will suggest the magnitude of the visual difference. Comparing or transferring color temperature numbers between different applications is not something I would recommend doing.

Here's some more background on the determination of the CCT:
http://en.wikipedia.org/wiki/Planckian_locus#Correlated_color_temperature

Cheers,
Bart