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

  • From: Graeme Gill <graeme@xxxxxxxxxxxxx>
  • To: argyllcms@xxxxxxxxxxxxx
  • Date: Mon, 24 Dec 2007 01:09:45 +1100

lee scratchy wrote:

actually I was more wondering about the DE measurements precision on the s2
> compared to the i1 display(which is supposedly very close to the i1 pro).

when the s2 gives me a 0.8 average DE, does that mean it's completely off..
> ..and might just be >10 average if I was to check with a i1 display ?

You can't tell what the distribution of errors is from the average. 0.8 DE
average is generally something that would be regarded as good. But it
says nothing about the absolute accuracy.

but you seem to agree that the s2 is quite off on that point ? also with LCD ?

It's impossible to say without a reference instrument to compare it to.
The self check error reports indicate how well the display has been
made to conform to the target, and how consistent the measurements
are. But the absolute accuracy can't really be determined without
something independent to compare with.

I'm already VERY happy with the PQ, but the idea of achieving better DE with a 
i1
> display is starting to grow on me ;)

It may be slightly more consistent in its readings, and therefore
give a lower self check report, but the visible effect would be
hard to predict. Each type of colorimetric instrument will probably
have a slightly different calibration for a given type of display,
and therefore if compared side by side, may look slightly
different. Once again, compared in isolation you may be
hard pressed to see a difference due to this cause.

And just to correct myself:

The main advantage of the Eye One Pro is that it is going to be more
accurate with any spectrum that a colorimeter has been specially
                                                    ^ not
calibrated to. The disadvantage is that it tends to collect less
light than a colorimeter, so you really need a longer integration time
to give a repeatable reading.

Graeme Gill.

Other related posts: