ACR/Fors Calibration skin tone problem

[Rick McCleary]

I have run the Fors calibration script (http://fors.net/chromoholics/) on both of my cameras (Canon 5D, Canon 20D) under very controlled conditions (studio strobe light).  

I have noticed a persistent problem and wonder if anyone else is seeing it:
When using the Fors calibration, greens and blues become much cleaner and richer, but skin tones become HOT.  A fair amout of correction is required.  Simple curve-based color correction does not do the job; for best results, I have to Apply Image to the red channel from the green channel.  This syndrome becomes a real problem when separating to CMYK.  For good CMYK dimensionality in skin tones, there must be a fair amount of cyan - 20-30% of the magenta value.  If the cyan is too low (or non-existent), the skin tones become too hot and somewhat flat.

Currently, I am using two fixes:
1) In the ACR calibrate tab, I'm deviating from the Fors calibration numbers significantly.  I'm adding 10 points to the Red Hue (which actually shifts to cyan), subtracting 4 points from the Green Hue (which shifts to green - away from magenta), and adding 10 points to the Blue Hue (which shifts to yellow).   While this does not totally fix the problem, it gets me in the ballpark to create very rich and dimensional skin tones.
2) Use Capture One :-).  I imagine I'll switch to C1 entirely once they incorporate DNG.

In the meantime, I'm trying to optimize ACR.

Any other Fors people out there who are experiencing this?

Rick

[danaltick]

Rick,

I just ran the ACR Calibration last weekend on a controlled shot I took of the color checker.  Everything seems fine to me.  Here are the results I got for my 20D:



I'm assuming that you are white balancing your images in ACR before looking for color problems.

Try keying in the calibration values and let me know how it looks to you.

Thanks,
Dan

[Rick McCleary]

Dan,

I'm running the script the same way.  The resulting calibration settings give a great rendition of the color checker, but need quite a bit of  tweaking when it comes to real life :-).

[danaltick]

Rick,

At this point, I'm not really sure what the problem is.  If you find it is a systematic problem, please post it here.  I'm not using CMYK and I haven't noticed any off color problems with my images since calibration, but if I use the "As Shot" for the white balance, the images do look too red most of the time; therefore, I have it set to Auto as my default.  Once the white balance has been corrected though the colors look good.  I will be the first to admit though, I haven't studied the input side of color management as extensively as the workflow and output sides, so I may be overlooking something here.

Dan

[Michael_S]

Rick,
I shoot a Nikon D50 so naturally I'm in a different world from you.  Although I really am smitten by the Canon 5D.  Anyway.....

I have used the Fors script and done ACR color calibration by hand in three different lighting conditions (cloudy mid-afternoon, sunny mid-day, and indoor with direct flash).  So far, I have not seen the poor performance in skin tones as you are getting  But I have ongoing concerns that the reds are a little too powerful now.  But each time I go get the object and look at it again it appears to be a close representation of the actual color.  However, I'm still watching it carefully.

I'm not sure how applicable my results are to you.  Probably not very useful at all.  But, for what it's worth, I wanted you to know that your concerns were being taken seriously by others in the forum community.

Best Regards,
Michael

[alanackoff]

Rick,

I have the same dissatisfaction as you with ACR with and without the Fors tweeks. I don't know why others don't see the same thing but there are so many variables (such as monitor model, calibration, and profiling) that we can't assume we are all viewing the same thing. Also I think color perception in the photo world is akin to pitch perception in the music world. Some folks have absolute pitch, some have relative pitch, and some are clueless.

FWIW I'm getting very nice skin tones with Capture One and the Etcetera camera profiles, Canon DPP, and Irident's Raw Developer. IMHO skin tones are a weakness in ACR and Lightroom. Unfortunately this puts me out in the cold as far as adopting the DNG based DAM methods recommended in Peter's book. At this time his workflow is of necessity Adobe-centric. I expect Capture One will support DNG files in the future--they have said that they will.

I've seen a post by Chuck Westfall of Canon indicating that there are issues pertaining to proprietary raw formats. I assume there is a "secret sauce" of some kind. I've used Canon DPP enough to think it's possible. DPP has an uncanny ability to render raw files almost perfectly and with virtually no tweeking which is matched by no other software. With tweeking, I can do a little better with some of the other converters, but have never been able to get first rate skin tones with ACR. Love the ACR workflow so it's frustrating to have to choose between image quality and convenience. I guess if you are building a raw converter to work with every camera on the planet there are bound to be compromises.  For me, image quality is the first priority but YMMV.

[BobSmith]

I use the Fors calibration on a Kodak 14nx 760, and 720x; and a Canon 5D and 30D (minor hacking involved to get it to work on a 30D).  I'm quite pleased with the results... especially the consistency of look across shots of the same subject with from very different cameras.  I've not seen the hot skin tones that you describe.  I'm not sure where the error you're seeing might be coming from.

Bob Smith