I can't resist commenting that the extensive quoting of Dr. Brill is the most coverage Datacolor has received on this list in a long while. As a friend and colleague of Mike I respect his comments, but I don't believe his proposal is the one end users are looking for. Actually, describing the needs and views of the end user to Mike is one of my main roles with him. My suggestion would be that the gamut of a color capture device be considered "the contiguous zone, outside of which the device cannot continue to distinguish color differences in a meaningful way", specifically further increases in saturation. If a capture device can continue to effectively distinguish color without changing settings on the device, then it's "gamut limit" has not yet been reached. This simple concept can then be appended with most everything else that as been said here, but the end user is unlikely to take away much from these further complications. Color scientists, on the other hand, will. C. David Tobie Global Product Technology Manager Imaging Color Solutions Datacolor inc. cdtobie@datacolor.com www.datacolor.com On Nov 7, 2011, at 5:03 PM, Richard Wagner <Rich@WildNaturePhotos.com> wrote:
On Nov 6, 2011, at : Iliah Borg <ib@pochtar.com> wrote:
I think we have a little language problem here. To quote, "there is no such thing as a camera, or scanner, gamut" http://www.cis.rit.edu/research/mcsl2/outreach/faq.php?catnum=3#255
Thanks for the link, Iliah.
"there is no such thing as a camera, or scanner, gamut"
There are several interesting discussions of this postulate in the literature. Unfortunately, the articles are expensive unless you have access to a research library, so I've taken the liberty of summarizing two of them for those without such access.
Berns (1) takes the above position and argues that the term ‘‘color gamut’’ historically has been associated with color output such as optimal color stimuli and additive and subtractive imaging systems. He notes that from both historical and current contexts, ‘‘color gamut’’ is associated with producing color, not measuring color; that is, color gamuts are defined by output devices, not input devices. If a camera is considered a measurement device - not unlike a colorimeter or spectrophotometer - it cannot have a "color gamut" as it is able to respond to all colors presented to it (i.e., there is no such thing as an "out of gamut" color for the device).
Berns takes issue with the use of the expression "color gamut" in scientific comparisons of digital cameras, and argues that "if a measurement device has a color gamut, ... [then] a miscalibrated spectrophotometer (calibrating to black instead of white) or camera with lens cap attached have small color gamuts." Rather than use the term "color gamut," he proposes the use of the expression, "color gamut rendering."
"That is, under a set of stated conditions a target of colors has been rendered to a set of colorimetric coordinates owing to various parameters. The stated conditions might include an ensemble of spectral reflectance factors obtained using a spectrophotometer with stated measurement geometry (the target), the camera-taking light source, the camera’s optical properties and methodology of obtaining colorimetric coordinates if known, and the standard illuminant and observer for the colorimetric data. Differences in input device characteristics would result in the target’s color gamut being rendered differently."
Brill (2) urges the explicit adoption of a definition for the camera color gamut. In his view, a viable definition seems to be the set of CIE XYZ triplets such that, for any such XYZ, there exists a spectral power distribution in a predefined set whose camera values map through a predefined algorithm to that XYZ.
Brill notes that the term "color gamut" is problematic when applied to a scanner or digital camera. However, he acknowledges that it has been used in this context in various venues, including referred articles and conferences. In such discussions, "the definition is always assumed and never stated." Brill believes it is time to "make explicit what is implicitly understood, or metaphorically, to spray-paint the invisible definition."
Brill discusses alternative definitions: he finds it "hard to grasp what could be a camera color gamut, because any light (visible or not) maps to some point in the camera color space—even if it is to the origin of the space. One could define a sensor (camera) gamut as the set of spectral radiances to which the sensor responds, but that would equate ‘‘spectrum’’ to ‘‘color’’ and the gamut could not be characterized in a space of reduced dimensionality such as a color space. Alternatively, a camera gamut could be ‘‘the largest set of scene colors that produce unique outputs,’’ but this definition also has difficulties." Brill notes that the term ‘‘camera color gamut’’ is a misnomer, because what is being discussed is not a property of just a camera. Rather, it is a property of a set of spectral power distributions (A), a camera (B), and an algorithm that maps camera values to CIE XYZs (C).
Brill ultimately proposes the following definition:
"Camera color analysis gamut: Attribute of a set of spectral power distributions (SPDs) (A), a camera (B), and an algorithm (C) that maps camera values to CIE XYZ triplets, that attribute being the set of CIE XYZ triplets for each of which there is an SPD in (A) that, when acquired by camera (B), produces camera values that, when processed by algorithm (C), produces that XYZ triplet. The gamut can be represented in a color space derived from XYZ triplets (such as CIELAB), or in a reduced space such as chromaticity (x,y). (The alternative term camera color gamut is also used). "
Brill closes with the following statement: A definition of camera gamut is new, and needed. The definition I propose here is now being discussed by representatives of several standards bodies. The quality of the algorithm, or of the camera, or of the set of SPDs, are valid subjects of research papers, but one must start with a definition such as the one here so people know what is being discussed.
Both of these short papers are excellent and make points worth understanding and remembering.
--Rich Wagner
===
(1) Let's call it “color-gamut rendering” COLOR RESEARCH & APPLICATION Volume 32, Issue 4, August 2007, Pages: 334–335, Roy S. Berns
(2) Camera color gamut: Spray-painting the invisible definition; COLOR RESEARCH & APPLICATION; Volume 32; Issue 3; June 2007; Pages: 236–237; Michael H. Brill
_______________________________________________ Do not post admin requests to the list. They will be ignored. Colorsync-users mailing list (Colorsync-users@lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/colorsync-users/cdtobie%40mac.com
This email sent to cdtobie@mac.com