[argyllcms] Re: Compatibility with i1 (Eye One) Display Pro 3 retail and OEM?

  • From: János, Tóth F. <janos666@xxxxxxxxxx>
  • To: argyllcms@xxxxxxxxxxxxx
  • Date: Tue, 14 Feb 2012 20:37:09 +0100

I use the terms valid and invalid when I speak about theories, like a
good and wrong solution for a math equation, like how a matrix profile
is valid for an additive display but invalid for those which mix the
colors „randomly”.

Unfortunately I can't compare any of them to a lab spectro.
What I noticed is that my CM and a borrowed i1Pro spectro agree nicely
with the built-in colorimeter of a CG275W (I get dE2000 below 1.5 if I
validate the hardware calibration with those external sensors) while
the i1d3 with factory EDR (which was built for this IPS LCD panel by
X-Rite) or customly created ccss highly disagrees (dE2000 around 7.5).

I would like to keep the dE2000 below 4

What I don't like is that I can't use a ccmx file created with 6500K
white point and set the white to 5000K with that colorimeter and
measure back ~5000K with the spectro. The colorimeter reads 5000K when
two spectros (which agree nicely) say it's smething else (regardless
if all the three agreed on 6500K - using the ccmx made on 6500K for
the colorimeter).

2012/2/14, Gerhard Fuernkranz <nospam456@xxxxxx>:
> Am 13.02.2012 22:31, schrieb János, Tóth F.:
>> I just missed the point that the ccmx seems to be WP relative and thus I
>> can't use a profiled i1d3 to set up the white point or create profiles if
>> the display is not additive.
> Hi Janos,
> you can use any instrument to set a white point - but WITHIN THE UNCERTAINTY
> LIMITS of the instrument (or of the measurement workflow).
> Without ccmx this is the native uncertainty of the i1d3.
> With ccmx, the resulting uncertainty is basically the sum of the uncertainty
> of the reference instrument used to create the ccmx + the residual error of
> the correction matrix model + repeatability error of the i1d3.
>> I thouht this problem will be solved with ccss corrections but it doesn't
>> really work well for my i1d3. I imagined a profiled i1d3 will be as
>> accurate as the spectro when I use it on the display I profiled it for.
>> But it's not.
> Well, you notice a "disagreement" between the two instruments, but how can
> you know which of them is actually more accurate, w/o comparing to an even
> more accurate reference instrument? Couldn't the spectrometer readings be as
> far off from the (unknown) "true value" as the i1d3 readings?
>> The ccss isn't good enough and the ccmx becomes invalid as soon as I
>> adjust the white point to a different color temperature or I measure a
>> different gradation level where the color mixture is different. And it
>> also seems to be inaccurate on non-additive displays...
> You are using terms like "valid", "invalid", "good enough". But how good is
> "good enough"? Or where is your threshold between valid and invalid? Can you
> quantify your accuracy expectations? For instance, +/- how many (x,y)
> chromaticity units of measurement uncertainty are you willing to accept?
> Best Regards,
> Gerhard

Other related posts: