Color difference equations [was: Color Measurement]
Color difference equations [was: Color Measurement]
- Subject: Color difference equations [was: Color Measurement]
- From: Marco Ugolini <email@hidden>
- Date: Wed, 15 Jul 2009 22:04:04 -0700
- Thread-topic: Color difference equations [was: Color Measurement]
In a message dated 7/15/09 8:12 AM, Mike Eddington wrote:
> 2.increase your tolerances to account for instrument differences, or
> perhaps use dE2000
Mmmm...I find it interesting that the DeltaE 2000 color difference formula
seems to be considered by many as a way to "cheat tolerances".
I've been surprised to see lately, coming from different quarters, (and I
freely paraphrase here, so don't take my description too literally) the idea
that the DeltaE 76 color difference formula is the "tough one", the one that
produces big numbers which force one to be manly and try to tame the big
beast of color matching -- whereas DeltaE 2000 is for the faint-hearted, for
those who cannot handle the cattle-rustling realities of DeltaE 76.
You may or not take exception with my description (and I do invite others to
chime in with their own view), but I do find this way of approaching the
subject of color difference formulas a bit silly. Personally, and don't get
too upset if I say this, I don't think that DeltaE 76 is good for much any
longer. Color difference formulas have moved too far forward for us to
remain stuck with the coarser math of DeltaE 76.
As Roy Berns points out in "Billmeyer and Saltzman's Principles of Color
Technology" (in particular in Chapter 4), in order to reflect the
psychophysical qualities of human visual perception, color difference
equations need to be "optimized", meaning that they must transform the
measurement data nonlinearly to reflect a series of psychophysical
principles (he mentions the "non-Euclidean" nature of color space to
illustrate what he means).
In that light, DeltaE 76 strikes one as a very simple linear formula, as
such very much Euclidean:
DeltaE ab = sqrt { (L*_2-L*_1)^2 + (a*_2-a*_1)^2 + (b*_2-b*_1)^2 }
Too simple, in effect, since it does not include any modifying and
multiplying factors which attempt to reflect the nonlinearities of human
visual perception within the CIELAB space depending on lightness,
chromaticity and hue -- not to mention other complicating factors like
geometric components (surface finishes like gloss and and matte, for
example), or the effects of surrounding colors on the perceived sample, and
so on which are the subject of more-recent studies like the ones on color
appearance models. (But I digress.)
DeltaE 2000, however imperfect it still is, includes a number of modifiers,
corrections and multipliers that attempt to account for perceptual
nonlinearities and lack of uniformity, as reflected by the formula's
startling complexity when compared to the bare-bones of DeltaE 76:
<http://en.wikipedia.org/wiki/Color_difference#CIEDE2000>
To make a long story short, I am not a color scientist, but, within my
limited understanding of the subject, I think that holding on to a view of
DeltaE 76 as a formula fit for the "tough" ones amongst us is misleading,
and has more to do with a misplaced machismo than with its effective
usefulness when compared to far more discerning, and I think practically
useful, formulas such as DeltaE 2000.
Marco Ugolini
_______________________________________________
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