Re: Color difference equations [was: Color Measurement]
Re: Color difference equations [was: Color Measurement]
- Subject: Re: Color difference equations [was: Color Measurement]
- From: Klaus Karcher <email@hidden>
- Date: Fri, 17 Jul 2009 09:11:38 +0200
Hi Steve,
Your objection is completely justified IMHO: measuring color means
measuring perception and if our measure doesn't correlate with our
perception it's simply inadequate.
Steve Upton wrote:
While dE76 can be an effective process control tool, if we don't
ensure that the differencing correlates to perception then we might
as well be recommending good ol' density.
... or even purely radiometric units to take it to extremes ...
Isn't the reason we want to
use colorimetry in process control that we want to bring things
closer to perception?
It is undisputed that there are situations where more physical measures
are appropriate (i.e. when we want to spot the /sources/ of errors), but
when we are interested in the /effects/ of deviations to our perception,
we should use the measure that correlates with our perception at best --
and that's CIEDE00 (at least for small color differences).
The only issue I have with CIEDE00 is that we can't decompose the
distance into perceptually independent components like Hue, Chroma,
Lightness, Redness/Greenness and Blueness/Yellowness.
It's a pity that the DIN 99 colorspace
<http://www.engl.dfwg.de/doc/dfwg homepage engl-499.htm> never
gained the popularity it deserved as it is a simple to use euclidean
colorspace with much better correlation to the perception of small color
differences than CIELAB.
In the meantime there are several attempts to bring together the
advantages of CIEDE00 and CIELAB (e.g
<http://www.art-si.org/PDFs/Processing/JOSA2007-Embedding-nonEuclidean-Urban.pdf>)
or to deduce decomposable color difference metrics from color appearance
models like CIECAM02.
Then there's the issue of dE2000 doing a better job of
"democratizing" the color space. Metrics and their tolerances (like
Max dE) tend to be applied in the same way throughout the color
space. But if dE in yellow is a very different commodity than dE in
gray then Max dE becomes a color-dependent quantity which it is not
supposed to be!
unless I'm missing something...
Klaus
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Colorsync-users mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden