[argyllcms] Re: Monitor profiling questions

Michael Grigoriev wrote:

There are several weird things. I read that a lot of people seem to have
trouble getting their LCDs to be bright enough for the target of ~140
cd/cm^2 - my problem is the opposite! I can't make my monitor dim enough to
hit 140. After the monitor warms up, even at brightness 0 the luminance
reads ~150. Is that normal?


It probably depends on the display, and the nature of the brightness control.
The desired brightness is usually a function of the work environment - it's
desirable that the brightness of the surrounding environment be similar
to that of the display. Some people have other criteria, such as matching
the brightness of prints in a viewing booth, or matching some industry standard,
or matching several different monitors.


But the main problem is that the profile created at the end of the process
seems rather wrong. For one thing, the avg error reported at the end of
profile creation is ~80 for the matrix/shaper profile and ~40 for LUT
profile. That seems awfully high....\


Such numbers usually mean that the profile is completely broken. Typically
this happens if the device values and reading get mixed up, or something 
similar.


The profile looks wrong too. The blacks are rather light compared using no
profile. I thought about this for a while, and I suspect it could be caused
by another abnormality I noticed when calibrating the black point of the
monitor. It reads very high on blue - even if I turn down the blue gun all
the way to zero and red and green to 100, it tells me I should be decreasing
the blue. Could it be that the backlight of the LCD has a blue tint, and so
the profile has to bring up red and green to compensate for it, thus losing
dark color saturation? Does that sound possible? Any way to avoid that?


LCD's are known for having strange behavior at the black end, because unlike 
CRT's,
the black point can be quite different to the white point. By default dispcal 
will
try and calibrate with a uniform grey balance from black to white, but this may
be asking too much of some LCD's. If this is the case, you can use the -k option
to change the target, for instance -k 0 will make the native black point the
target for the grey balance, and it will blend from the native black point
to whatever the white point target is. You will want to set the RGB guns 
controls
back to their defaults before running with the option.


Also, overall, the profile seems to have a bit of an orange skew to me, but
that could actually be correct - it's possible that I'm simply used to
looking at a colder colors.


See how you go fixing the black end calibration, and then re-do the profile
readings. If you still get a poor profile fit, run profcheck on the .ti3 and 
the .icm,
using -v, and see what sort of discrepancies there are. If some of them are
very large (ie. delta E 100+), then this indicates a severe problem in samples
being mismatched somehow. I'm not sure quite how that would happen with a 
display
though. Send me the .ti3, and I'll take a look at it. One would usually expect
a profile fit average delta E in the 1-3 range.

Graeme Gill.

Other related posts: