[argyllcms] Re: dispcal: Warning - RegOpenKeyEx failed with 2

  • From: Gerhard Fuernkranz <nospam456@xxxxxx>
  • To: argyllcms@xxxxxxxxxxxxx
  • Date: Thu, 03 Jan 2008 17:26:40 +0100

Graeme Gill wrote:
I can't say I really understand what the HCFR folks are doing with their reference
files. It doesn't make any sense to me that they can "calibrate" an instrument
without an independent reference of some sort. So what I've done with
my driver is to hard code a reference of the particular HCFR I have
into the software. The reference was created by calibrating against
an X-Rite DTP94 on two particular displays. The validity of this calibration
depends greatly on the uniformity of manufacture of the optical sensor,
and the similarity of the spectral characteristics of the displays to
what is actually being measured. Without some sort of manufacturers
factory calibration for each HCFR instrument, I don't see what
else can be done for it.

This study was done with MAZeT's MTCS-C2 Colorimeter Board, which even contains sensors with a spectral sensitivity approximating the CIE standard observer, and each board is also factory calibrated (I don't know for which spectrum, but I would at least assume that they are using a lab-grade reference to do the factory calibration). Nevertheless the study show that the factory calibration result in errors of up to 10 delta E when measuring CRT and LCD displays. Only with an individual calibration to the monitor's spectrum, the error could be reduced to about 1 delta E.

So if even approximately colorimetric sensors with an individual let me call it "general-purpose factory calibration" don't work sufficiently well for measuring a display device, I don't want to image how non-colorimetric sensors w/o any factory calibration will work...


Other related posts: