[argyllcms] Re: adjusting monitor settings
- From: Michel Joly de Lotbinière <michel@xxxxxxxxxxxxxx>
- To: argyllcms@xxxxxxxxxxxxx
- Date: Thu, 13 May 2010 16:11:41 -0400
I doubt there is any simple, quick method of determining the best
initial hardware adjustments--after all, it's the whole chain of
calibration+profiling that determines their validity in terms of
colour reproduction.
If I had the LCD controls, I'd simply adjust them with using the
various initial set-up measurements in dispcal until I achieved the
closest match to target black/white brightness levels and colour
temperature with the least possible reduction on the native contrast
ratio. The calibration can handle everything else.
Grey scale quality and tone neutrality are the big wins of a properly
calibrated+profiled monitor, simply because if those are good,
everything else in our perception of colour reproduced on an LCD seems
much more pleasing.
As a rank beginner, and using an LCD that only offers a brightness
control, I use Florian Höch's dispcalGUI, simply because it allows
easy unattended monitor profiling while offering all the adjustments
available in the various command line tools. It speeds the entire
ArgyllCMS process.
But there is another advantage in recent versions: an integrated
profile verification/testing routine allows rapid, repeatable tests to
be done to compare the results of profiling settings. There's a
bundled a 54-patch ti1 test file that includes a grey scale (but of
course you can generate and use your own ti1 file), and the whole
procedure produces a nice HTML page with the various calculated
overall deltaE, as well as the error for the individual patches.
Inaccuracy of the profile in particular parts of the gamut will be
obvious.
It's really helpful to have some objective comparable numbers when
fiddling around with the profiling options, and doubly so to see their
effect on the error of individual colours and the grey scale. I was
able to achieve a profile with much more neutral greys and lower
errors on other colours by specifying the target black and white
brightness to be the same as that actually measured on the screen and
opting for partial correction of black point hue (-k=0.5). Visual
comparison of profile results only worked for the more extreme
adjustments (and vision is constantly compensating and adjusting
anyway), the measured numbers are much more useful to zero in on
correct parameters.
For instance, I was able to achieve:
Whitepoint ΔE*76 0.26
Whitepoint ΔE*94 0.19
Whitepoint ΔE*00 0.22
Average ΔE*76 0.43
Average ΔE*94 0.33
Average ΔE*00 0.34
Maximum ΔE*76 1.2
Maximum ΔE*94 0.78
Maximum ΔE*00 1.05
Median ΔE*76 0.41
Median ΔE*94 0.33
Median ΔE*00 0.33
Median absolute deviation ΔE*76 0.2
Median absolute deviation ΔE*94 0.14
Median absolute deviation ΔE*00 0.14
Standard deviation ΔE*76 0.27
Standard deviation ΔE*94 0.19
Standard deviation ΔE*00 0.21
RGB gray balance average ΔC*76 -0.06
RGB gray balance maximum ΔC*76 0.41
on the 54-patch ti1 set.
Of course, the joke is reproducing these results on an average day
with my average LCD, and the outcome is never very serious: just now I
tried the "same" measurements, and everything came out with about 3X
those statistical quantities. C'est la vie.
I'm not sure if I answered your question....
Other related posts:
