[argyllcms] Re: ColorMunki measurement drift

  • From: Juergen Lilien <jp.lilien@xxxxxxxxxxxxxx>
  • To: argyllcms@xxxxxxxxxxxxx
  • Date: Sun, 26 Sep 2010 17:19:06 +0200

Roger Breton wrote:

Hello Roger!

> I wish you had an EyeOnePro to compare, ...

Yeah, that's my wish too. ;-)

> ...because, to my knowledge, EyeOnePros are not susceptible to the
> effects you describe (if EyeOnePro drifted as much as you report, I would
> start to worry).

I started with a Spyder2 years ago, but the results were not to
my satisfaction, so I thought to make a big step forward with
the ColorMunki, only to find there is still something fishy
with my instrument readings.
It would be really interesting to compare with an EyeOnePro, but
I've read a recommendation to sent an EyeOnePro once a year to
the manufacturer for recalibration, so I would expect that inter
instrument agreement is not without problems, especially between
different types of instruments (e.g. EyeOnePro and ColorMunki).

Of cause, I would by happy with my ColorMunki to get very similar
results with two measurements being 30min apart.

> Do you leave the ColorMunki hung to the screen for an hour before making
> your measurements? To try minimizing thermochromism effects?
> It should help, in my opinion.

Actually I found the problem during the interactive calibration
of the monitor. It is really a hunt to set the white point within
<0,2 DE. After several readings the DE starts to increase again,
so I have to push the monitor controls one more step, and so on.
I found that this iterative procedure takes more than 30min to
get in the region of more stable readings, that actually result
in better looking LUT curves. So yes, recently I leave the ColorMunki
roughly an hour on the screen (measuring the white point) before
I start over with the real calibration/profilation.

But the problem remains, that the agreement between monitor (preset)
and ColorMunki constantly degrades after the first reading (without
acclimatization time) even with self-calibration.

E.g. monitor preset is 6500K (Blackbody):

First measurement of uncalibrated response with "cold" ColorMunki:

Black level = 0.22 cd/m^2
White level = 156.70 cd/m^2
Aprox. gamma = 2.19
Contrast ratio = 721:1
White chromaticity coordinates 0.3115, 0.3242
White    Correlated Color Temperature = 6610K, DE 2K to locus =  2.1
White Correlated Daylight Temperature = 6612K, DE 2K to locus =  2.8
White        Visual Color Temperature = 6534K, DE 2K to locus =  2.0
White     Visual Daylight Temperature = 6720K, DE 2K to locus =  2.7

This looks quite reasonable, but 19 consecutive measurements
_with_prior_self-calibration_ later I get this:

Black level = 0.24 cd/m^2
White level = 152.46 cd/m^2
Aprox. gamma = 2.17
Contrast ratio = 642:1
White chromaticity coordinates 0.3064, 0.3196
White    Correlated Color Temperature = 6958K, DE 2K to locus =  2.4
White Correlated Daylight Temperature = 6960K, DE 2K to locus =  2.4
White        Visual Color Temperature = 6854K, DE 2K to locus =  2.3
White     Visual Daylight Temperature = 7065K, DE 2K to locus =  2.3

Despite the self-calibration the black level is up, white level + gamma
(+ contrast ratio) are all down and the white point has changed about
nearly 350K in the wrong direction.

What's going on here?

Best regards, Juergen

Other related posts: