There are many ways to get a gamut size, and they are ALL
approximate. What are two different colors? Two colors
separated with a Delta-E* of one? ...or whatever other
criteria? And is it Delta-E* CIELAB or Delta-E* CIE94, or
Delta-E* CIEDE2000?
Bruce Lindbloom shows a comparative chart of the L*a*b*
gamut efficiency, which indicates the percentage of the
visible color volume contained in various RGB spaces.
According to his calculations, Adobe(1998) contains 50,6%
of the L*a*b* gamut, and sRGB 35.0%. So Adobe RGB is
roughly 50% bigger than sRGB. Your sRGB number of 820,000
and 1,2 million for Adobe are in similar proportions.
However, I would be VERY careful in assigning an ABSOLUTE
number of colors in each case. It is well known that the
L*a*b* color space is not totally uniform. The CMC, CIE94
and CIEDE2000 color difference formulas were developed to
"fix" that problem, but they do it by applying varying
weights on L*, C*, and H* depending on the color location
in L*a*b* space. This is the same space used, without
weights, in determining the gamut efficiency, so the gamut
efficiency is still an approximation when used to compare
the number of colors!
In a 1998 comment to the editor in "COLOR research and
application" (Volume 23, Number 5, October 1998", p337)
Mr. C.S. McCamy, which is more than well known in the
color field (think ColorChecker), says:
"Knowing what they did about color judgments, they were
able to conclude that, under usual conditions, a normal
observer could distinguish 1,875,000 colors and, under the
best viewing conditions, 7,500,000 colors. "
The "They" in his sentence are:
Dorothy Nickerson and Sidney M. Newhall, A psychological
color solid, J. Opt. Soc. Am. 33, 419422 (1943)
His comments were made in regards with the number of 2.28
millions colors presented in the article by:
M. R. Pointer and G. G. Attridge, The number of
discernible colours, Color Res. Appl. 23, 5254 (1998).
From the comments and the comments reaction by M.
Pointer, it seems there is "kind of an agreement" for a
number around 2 millions for "usual conditions", and this
number will vary significantly with the viewing
conditions.
Thus, two points:
1- You are better using relative ratios than absolute
numbers. If absolute numbers of colors the only info
available, use the same program for comparison. If the
ratio of the absolute numbers of colors do not match the
relative ratio as computed with the L*a*b volume, you
better double-check and ask how they did their
computation.
2- Based on a rough average of 2 millions colors in "usual
condition" (conditions which need to be matched!) you
could estimate, from the gamut efficiency numbers given
above, the numbers of colors in Adobe RGB at 1.1 million,
and 700,000 for sRGB. Multiply these numbers by four for
"Best" conditions!
My conclusion is that using absolute number of colors will
more often result in conflictual values, since any
reference number can be selected for the total number of
colors, whereas using L*a*b* volume ratios is better, even
if colors are distributed in a somewhat non-linear fashion
within it.
Danny Pascale
email@hidden
www.BabelColor.com
On Sun, 18 Feb 2007 19:30:12 +0100
"Olivier Desmaison" <email@hidden> wrote:
There are many references that indicate that sRGB has a
gamut volume of 820,000 colors and AdobeRGB about 1,2
million. If I check a good RGB glossy output profile, it
comes close to 1,4 million shown in MonacoProfiler. The
same plotted on an xyY 2D graph shows some consistency.
In a 3D graph the story is totally different : the
output profile is far smaller. I'm on XP but it seems
that on some other systems, Gamut Work from Monaco shows
that AdobeRGB is about 2.5 million, and sRGB 1.4, this is
also where I got the output profile volume.
I assume this is a gamut volume calculation issue, but I
don't understand.
Any help ?
Olivier
_______________________________________________
Do not post admin requests to the list. They will be
ignored.
Colorsync-users mailing list
(email@hidden)
Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/colorsync-users/email@hidden
