I bet, most of us find little or no use for ACR calibration tab in Lightroom or Photoshop. The settings there look really cryptic, and experimenting on setting them manually gives weird or unusable results. However, there are some software and photoshop scriptsm which can make ACR calibration meaningfull, and results of its usage visible. One of such scripts can be found at http://www.fors.net/chromoholics/. It requires Photoshop CS3 to run, a Gretag Macbeth color checker or mini color checker card and about an hour to run on a mediocre PC. It can also be run on CS2, but there you should solve a problem of maling RAW files for newer cameras compatible with an obsolete ACR 3.7 - the best way is to convert them to DNG with no adjustments made.

I photographed the color checker card just to experiment with. The photo is slightly defocused in front, but this does not matter for calibration. Results of calibration left me stunned. Just watch:

http://www.domusrest.ru/somepics/_MG_0187.jpg

http://www.domusrest.ru/somepics/_MG_0187-2.jpg

http://www.domusrest.ru/somepics/_MG_0187_show.png

Top picture is RAW converted with default settings. It was taken with 580 EX II bounced off the ceileng, the color temperature was set straight 5200K. Note reds and blues.

The second picture is the result of converting the same RAW with only ACR calibration parameters set to that calculated with the calibration script noted above.

The third picture is a composite one. Top lefts of the patches are colors before calibrations, bottom rights are colors after the calibration. Good monitor will display you shocking difference. However, even puny laptop LCD makes difference in blues, reds, lime, purple and skin tones visible. There are no difference in grey patches at all.

What this means to our practical applications. Here is one (not bright, just to show) sample:

http://www.domusrest.ru/somepics/_MG_0189.jpg

http://www.domusrest.ru/somepics/_MG_0189-2.jpg

Again, top picture without calibration, bottom calibrated. Note skin tones, lime colors on pillow, soft reddish tones in stripes and hairs. The calibrated picture looks deeper, 3Dish, and subjectively delivers more dynamic range.

For sure, best results are achieved, when you calibrate your particular camera and lens setup for the particular shooting sitaution. But this sounds like utopia. Most lenses, while giving some color transmission difference, are similar in color reproduction. Some lenses do deliver strong color cast, but I suppose, most of us prefer to use something really neutral.

Theoretically, every sensor may have different color reprodiction. However, I presume, differences are insignisicant. Saying that, I believe, calibration settings can be shared between the cameras of the same model given similar shooting situation. Here they are:

Shadow tint: -1
Red Primaries:
Hue: -18
Saturation: 32
Green primaries:
Hue: -21
Saturation: 14
Blue primaries:
Hue: 11
Saturation: -24

The test photo was taken indoors, using 580 EX II bounced flash, WB 5200K. The lens was 24-105 f/4L. These shooting conditions represents the situation for 'Faithful' picture style quite similarly. Calibration settings obtained should be compatible with most indoor flash shots, and in some range of frequently used color temperatures for shots taken without flash. This calibration, however, CANNOT BE USED in severely different lighting: tungsten or outdoor in bright sun.

I suggest you try these settings with your regular shots taken with 40D. If you study the histogramm of the second photos set, you'll see that reds and greens are shifted slightly and exposed significantly less - overexposed reds is a very common problem in digital shots, and calibration can help resolving that. You may also notice subtle change in greys, and much better skin tones. Try this and your feedback is welcomed.

I am going to continue calibration experiments with different shooting conditions, and will publish the results, if anyone interested. Don't wait for outdoors setups by now - the weather is really dull in Moscow now.