Message
Re: [colorvision_group] Re: Ooops - D65 != D65 ???
2007-06-20 by CDTobie@aol.com
In a message dated 6/20/07 12:44:30 AM, uli@... writes:
Do you have a link for that document? I had a look, but couldn't find it.
It does not appear to have made the leap to our new website But here is a description of the transform options, from other notes:
Chromatic Adaptation Mode
When recording colorimetric values in a profile, XYZ values need to be Chromatically Adapted from their measured color space to the D50 color space that is used by the ICC profile. There are different Adaptation Algorithms that can be used to perform this computation.
Bradford
The Bradford transformation is recommended by the ICC Specification and other sources as being the best adaptation method to be used. This is the default value in ColorVision software.
XYZ Scaling
Many other monitor profiling applications use the XYZ Scaling algorithm for Chromatic Adaptation when creating ICC profiles.
None
Prior to OptiCAL version 3.7.6, colorimetric values stored in ICC profiles created by this application did not undergo Chromatic Adaptation. This option no longer offered in recent ColorVision software.
> Don't expect identical results, even from the same tool used twice. Expect
> somwhat larger variations from differing software, and larger still from
> differing hardware and software, but all should be within reason. If they are way
> off, then thats an actual problem, rather than a statistical variation.
I posted my measurement results already in Apple's ColorSync-Users list, but they might
be interesting for readers of this group, too, and I'd be interested in whether you would
qualify these results as "being within reason", or rather as "an actual problem":
Target value: 6500 K
Which can be met by an infinite number of xy pairs... as was explained much earlier in your thread.
Eye-One Display: 0.315 0.341 6316 K
Eye-One Pro: 0.314 0.333 6407 K
huey: 0.307 0.328 6836 K
Spyder2Pro: 0.313 0.325 6510 K
(0.311 0.321 6657 K with ICC v4; is the chromatic adaptation
transformation lossy?)
(Kelvin calculated by entering the xy results into Spyder's "user-defined whitepoint" dialog
panel)
All of these values are probably within the published accuracy specs of the devices involved. You seem to be looking for absolute numbers from end user devices. Even laboratory Minolta meters costing nearly as much as my car have an accuracy spec that would allow for about this much variation. You need to have some perspective on this issue: spend tens of thousands of dollars on two Minolta lab meters, measure the same screen with both of them, get results which vary by, say, 200 or 300 K... thats just the tolerances one gets in this field. Coverted from little x, little y values with variations in the hundredths, to K values that are hundreds different; it sounds big, but thats the math fooling you.
C. David Tobie
Product Technology Manager
ColorVision Business Unit
Datacolor Inc.
CDTobie@...
www.colorvision.com
**************************************
See what's free at http://www.aol.com.