[argyllcms] Re: EyeOne Pro Not Calibrating Properly

  • From: Graeme Gill <graeme@xxxxxxxxxxxxx>
  • To: argyllcms@xxxxxxxxxxxxx
  • Date: Fri, 05 Dec 2008 00:56:02 +1100

Timothy Hattenberger wrote:
I plugged in a recently re-certified Eye One and argyll recognized it fine. I then calibrate on the white tile, and proceed to setting the monitor luminance. I just calibrated this display last evening with an I12 and set it to ~124 cd/m2, but the I1 Pro measured it at approximately ~178cd/m2. So I used the EyeOne match software with the I1 Pro to see what was going on, and that measured properly (~125 cd/m2).

Which version of Argyll are you using ?

I think the problem is that when I do a calibration of the I1 Pro in EyeOne Match, it turns the light on and takes a reflectance measurement of the white tile. The light does NOT come on when calibrating using Argyll, which I think I can convince myself would lead to the artificially high measurements.

You are doing an emissive measurement, and since there is no
absolute emissive reference available, the only calibration
possible is black level calibration. The light will only
come on when doing a reflective calibration, where there is
an absolute reference available (the white tile).
EyeOne Match is doing  an (unnecessary) reflective
calibration, along with the dark calibration performed
for both emissive and reflective measurement, Argyll is not.

I tried calibrating on a dark surface and also with no calibration (-N) and got the same results. Any ideas? I pasted the dispcal command below and also the data Argyll reads for the instrument upon dispcal initialization.

Getting the absolute calibration right is tricky. It depends a lot
on the geometry (the direction of the light being measured, and
the sensitivity of the instrument to light from different directions).
I've seen different instruments disagree by += 50% when measuring
the same source of light. Certainly my i1d2 and i1pro's don't really
agree with each other, even when using Gretags drivers (see below).
The disparity might well be greater on LCD displays, since LCDs
have a much narrower emission angle, and the i1pro is much more
directional than the i1d2.

The current Argyll drivers return results that are (as far
as I can tell) identical in absolute calibration to
the results using the Gretag/X-Rite drivers, so there's not much
more I can do but rely on their factory calibration.
It's not clear to me whether EyeOne Match is second guessing
the factory calibration (it may be, particularly if it marries
a particular instrument to a particular display), and/or whether
you are using an older version of Argyll. If the former is the case,
then they will be in trouble if X-Rite get their act together and
start calibrating their i1d2's and i1pro's against the same standard :-)

With the instruments I have, measuring the same spot on my CRT
using the Argyll drivers:

        i1pro rev D:    58.20
        i1pro rev A:    52.70
        i2 disp 2       41.01
        DTP94           42.11

Using the gretag drivers:

        i1pro rev D:    57.72
        i1pro rev A:    52.37
        i2 disp 2       41.38

(Note that my display drifted a couple of cd/m2 over the course of
 the measurements.)

Using an LCD display:

Using the Argyll drivers:

        i1pro rev D:    258.56
        i1pro rev A:    243.40
        i2 disp 2       186.13
        DTP94           186.74

Using the gretag drivers:

        i1pro rev D:    262.70
        i1pro rev A:    244.34
        i2 disp 2       184.42

[ I did come across an interesting bug in investigating this, and
  it's a bug I suspect that the Gretag/X-Rite driver suffers from as well:

  The threshold that is used to detect when the sensor is over range (saturated)
  is when an average of 10 spectral bands are over a threshold of sensor
  full scale. In the case of peaky light sources (such as displays), it's
  perfectly possible for a measurement to be under this number, and
  yet for the measurement numbers to be quite measurably affected.
  So it turns out that my automatic switch of display integration time
  from 2.0 seconds to 1.0 second for bright displays isn't happening
  when it should, and measurements of display brightness's around 275 cd/m^2
  or over are being distorted. I will fix this by setting a zero tolerance
  for over-range, and allowing for 3 different display measurement
  integration times in the future. ]

Graeme Gill.

Other related posts: