White Balance - Heureka!

[Peter_DL]

Thank you for this info, Peter. - Would this magenta-green calibration of the camera not be automatically included in a custom camera ICC or DNG profile?

The DNG profile editor allows to do so in the Color Matrices pane (-> white balance calibration),
however, from what we can tell such deviating offset with Tint is not already considered by Adobe in the profiles as shipped.

Best regards, Peter

--

[tlooknbill]

What hasn't been addressed in this discussion that I'ld like touched upon is what any given white balance (aesthetically or technically accurate according to color science and spectra data) does to individual colors that aren't included as part of the color target used to produce a DNG profile or regular canned ones like Adobe Standard.

Take for instance this portrait I came across linked to in a Photo.net Lighting forum thread:

http://www.flickr.com/photos/vimages/8298075493/in/pool-1773207@N23/

Pinky caucasian skin syndrome using any number of flash or bright noon daylight. This happens quite often in similar shots among quite a few photographer's galleries I've come across online. And what seems to fix this in my shots that exhibit this is to add green to the white balance or use HSL to fix it but then the reds/oranges suffer for other image elements not exactly close to this skin color.

Just wonder if a study on perception of how much WB influences some colors over others has been conducted.

I realize another way to fix this is on an individual color by color basis tweaking the DNG profile recipe in Adobe DNG Profile Editor. That can get pretty cumbersome and sometimes when doing this it can introduce posterization if you get too refined the closer that pinky skin gets into other reds and oranges that look correct on other image elements.

There's also the issue of adaptation in whether one sees pinky skin appear orangish tan that has nothing to do with the appearance of WB but rather tweaks to the color table I've demonstrated in the two caucasian skin appearances between a custom dual illuminant profile and Adobe Standard on the same image below.

Here's an experiment...First look at the DNG Profile version for a while and note how much yellow appears in the skin compared to the pinky skin flickr portrait above, then close custom DNG Profile image and stare a bit at your 6500K display white a bit and then go back and first stare at the Adobe Standard version for a while and then quickly switch to the custom DNG Profile. For me when doing this, the DNG Profile now looks more pink than it did before.

 

[Hening Bettermann]

Hi

--this is not a reply to Tims post just above (#41), but to Peter_DL (#40) --

Peter I don't quite understand you. What I had in mind was: If the magenta offset is a constant for any particular camera, rather than something that changes during the day - will it not automatically be corrected if one makes a custom profile for that camera? Just any reasonable profile?

The manual of Raw Therapee (version 4.0.9, page 31) has a table for temp and tint values for neutral rendering for a number of cameras. The tint values range between about 0.9 to 1.1. I wonder how this relates to your figure of plus-minus 10.

Kind regards - Hening

[Peter_DL]

What I had in mind was: If the magenta offset is a constant for any particular camera, rather than something that changes during the day - will it not automatically be corrected if one makes a custom profile for that camera?

When running the Chart Wizard of the DNG Profile Editor, the ColorChecker chart as shot gets automatically white-balanced (based on the second brightest neutral patch i.e. light gray), before the Hue/Sat.-corrections are built per color patch. However, the Temp/Tint adjusts from this preceding white balance do not flow directly into the profile e.g. by ruling white balance calibration in the Color Matrices pane.

It would not necessarily be wise to have this auto-included at this stage, while the software cannot differentiate between the two effects:
a.)  the variance of natural occurring daylight perpendicular to the Daylight curve or Plackian locus (as shot).
b.)  a systematic offset with Tint introduced by the camera/sensor.

Anyway I was referring to DNG profiles and the DNG Profile Editor.
ACR does not take ICC/DCP profiles. No idea what related profiling software does in this respect.

The manual of Raw Therapee (version 4.0.9, page 31) has a table for temp and tint values for neutral rendering for a number of cameras. The tint values range between about 0.9 to 1.1. I wonder how this relates to your figure of plus-minus 10.

Interesting though !


Best regards, Peter

--

[Hening Bettermann]

Thank you. So an ICC camera profile (which I have in use) may be different.