Hardware: Pentax K10D, use DNG raw format.
Software: Pgotoshop CS3, ACR 4.5, Windows.
ACR allows me to select one of four profiles:
(1) ACR 4.4
(2) ACR 3.7
(4) Pentax 1.00
And results of this selection is dramatic. And strange, too.
I have one shot with XRtie ColorChecker.
I select "ACR 4.4" (default) profile and set white balance by second gray patch in raw (18% gray) with White Balance Tool. Color temperature is 8000 and tint is -20. Ok.
I select "ACR 3.7" profile, repeat WBT, and get same results. Ok.
I select "Pentax 1.00" profile, and get.... Temperature 11500K and Tint +75!
I know, that profiles change color rendering. It is ok.
But why these two (ACR 4.4 and 3.7 is almost the same) profiles change color rendering SO DIFFERENTLY? Color temperature is universal number, daylight is about 5500K, good flash is always 5500K, D65 light source is 6500K (6504K), etc.
I thought, that good profile should give good neutrals, and in case of controlled light, calculated (by gray patch) color temperature should correspond to light source.
I understand, that different profiles could render skin tones, cyans, shadows and highlights slightly differently, and there are many discussions, which profile is better. But 3500K difference is not "slight difference" and is not question like "I like skin tones with this profile more that ones with another"! Am I right?
It seems, that one of profiles is broken. Which one? Why difference is so huge?
Essentially embedded was broken, not in the colors, but in the white balance mappings. There was an issue with the white balance values in the profiles originally embedded in the Pentax DNG files. As long as you do a click-WB __AFTER__ choosing your profile, the actual colors should be fine (but won't and aren't expected to match the ACR 4.4/3.7 rendering exactly).