Subject: Re: ACR Calibration
Barry Clive Pearson wrote:
> Peter, you say "Batch deviations in fact seem to be a myth". Perhaps
there are enough people here using the same camera model to test this.
They could run the same script, on raw GM CC images from the same camera
model, and see how the results compare. <<
Yep, a round-robin test could be interesting. From my experience (as a chemist) I can say that it’s most critical to keep everything on an objective, measurable basis. And, the testing procedure should be as clear as possible, without seeking for senseless precision!.
Fwiw, below please find a proposal for a respective testing procedure. The core assumption is that we even don’t have to bring ACR calibration (and its variables) into the play in order to reach meaningful results:
1.) Starting point is a reasonably well-exposed capture of GM’s ColorChecker.
Full spectrum lighting via direct sunlight should be fine to minimize
metamerism. Also we have to consider that cameras change their gamut depending
on light conditions; that’s why ACR interpolates between two generic profiles.
Possible deviations resulting thereof should be kept at a minimum level.
So please nothing extraordinary like neonlight.
2.) Further I suggest to limit the comparison to 6 selected color patches.
Regarding the correct reference numbers, for the time being I suggest
to follow Bruce Fraser & Bruce Lindbloom. Their numbers for these patches
are +/- 1 the same:
red patch # 15: R 122, G 57, B 45
# 2 light skin: R 160, G 138, B 116
green patch #14: R 84, G 123, B 67
# 11 yellow green: R 143, G 168, B 73
blue patch # 13: R 59, G 49, B 126
# 3 blue sky: R 95, G 102, B 134
3.) ACR set to ProPhoto RGB (16 bit). Now, the key is to stay with Adobe’s
preset tone curve and the ‘native’ settings for Shadows, Brightness and
Contrast (ACR 2.x shipping defaults, 3.x without auto-correction). For
some cameras I saw the values 5/50/25, but I won’t claim that these are
generally valid. Further, Saturation slider at zero, Luminosity curvetool
linear and all Calibration tab sliders in their zero position.
4.) Whitebal the second gray, patch #20, via the eyedropper. This, btw,
should also eliminate any influence from the lenses and its coating.
5.) For every single color patch as listed above: *First*, adjust the
Exposure slider to match RGBmax to the reference value. For example for
the red patch, Exposure would have to be adjusted to get R = 122. Or,
for example for ‘yellow green’, Exposure would have to be adjusted to
get G = 168 (see above reference numbers). *Then*, note the 3 numbers
of the whole RGB triple.
That’s it. The resulting 18 numbers could / should shed some light on both aspects, ‘batch deviations’ and ‘compliance to a matrix space’ … imho.
Please let me know what you think.
P.S.: Above suggested analytical procedure is an excerpt of my manual calibration procedure. Insofar I’d be optimistic that it should work.
View All Messages in adobe.photoshop.camera.raw
ACR Calibration =>
Copyright © 2006 WatermarkFactory.com. All Rights Reserved.