scanners ,gamuts and urban myths
scanners ,gamuts and urban myths
- Subject: scanners ,gamuts and urban myths
- From: "tlianza" <email@hidden>
- Date: Tue, 26 Nov 2002 06:59:42 -0500
Hi to all,
I've been reading a number of questions and responses to this group so I
thought that I would chime in here. Strictly speaking, cameras and scanners
don't have "gamuts". If a color capture device has a minimum of three
overlapping, arbitrarily shaped color response functions whose spectral
range cover the visible, it, by definition encompasses the gamut of the
human visual process. Cameras and scanners with such wide span
sensitivities can sometime clip due to incorrect exposure or fluorescence in
the original scene, but they can capture and respond to every wavelength we
can see.
The problems start to enter the situation when we try to calibrate these
devices colorimetrically. In those cases, there can most definitely be
clipping and "gamut" limiting. If we calibrate these devices to an IT8
target, they are by definition, gamut limited.
A drum scanner, or negative film scanner, can be completely different.
These devices are generally NOT designed to capture color in general, but
the colors recorded on the film they have been designed to measure. In many
cases the sensors in these devices have much more narrow sensitivities. If
you visit the Kodak web site and look at the spectral sensitivities of the
sensors , you will see how they are manipulated based upon intended media.
In these products, it makes absolutely no sense at all to discuss gamut
because they are designed to measure film, not mimic eye response.
Digital cameras (at least at the High end) generally have sensitivities that
peak near the peaks in the CIE color matching functions. This gives them
some reasonable chance of producing predictable color. Once again, these
products can clip due to exposure or a high degree of fluorescence, but they
do not limit the gamut of capture.
The myth surrounding RGB workspaces and their effect on INPUT imaging is
based primarily on the theories and techniques used in managing the output
process. The fact is, on a digital camera, switching the workspace simply
applies a different matrix to the input data. If clipping occurs, it occurs
because someone wasn't careful in the design of the transformation, or they
used one of the ICC table based methods to profile the device. There is no
question that it makes sense to use one of the Wide Range spaces (Bruce,
Don, Adobe RGB, and RGB dejour) when MANIPULATING COLOR in Photoshop or some
other application. It makes absolutely no sense to use such a space in the
camera, unless you want to guarantee artificial problems. I would argue
that you are far better off using a relatively narrow color space definition
such as sRGB as a capture space. As a matter of fact for digital cameras, I
would recommend using a space that had a smaller defined gamut than sRGB and
a gamma much closer to 1 (say 1.8) than 2.2 . I would call this a
"transport" space. On the camera side it is just a mapping and there is a
lot said to allowing the user to boost saturation as needed. I was working
with a world class photographer on one of his books. It was a project
funded by National Geographic and the Discovery Channel. He had one
particular image that was driving the printer crazy. To it's credit,
Photoshop was showing the artifacts in the out of gamut region. It was one
of those hyper saturated Fuji Velvia images with a deep blue sky. Rather
than use the profile that was attached to the image, I had him tag it as an
sRGB image. I had him import the image and then locally correct the
saturation loss in regions other than the blue sky. Actually, he was so
hung up on the sky issues, that the other colors just looked fine to him
(even though they had been desaturated by this move....) . The image
printed just fine. I pointed out that the ROBUST management of out of gamut
colors is fundamentally impossible at the current state of the ICC
architecture. The combination of gamut decisions in the input profile and
gamut decisions in the output profile lead to some very bad artifacts. The
key was that the Hues were basically correct and it was easy to manually
shift the saturation around. One point that is often lost in all this
"gamut" speak is the fundamental priorities (in order of importance) in
subjective tone reproduction:
1. Correctness of overall HUE as adapted to the viewing conditions
2. Correctness of tone reproduction as adapted to the viewing conditions
3. Correctness of Colorfulness from subject to reproduction
4. Correctness of Luminance and Contrast range from subject to reproduction.
Usually, point 4 is impossible to meet, point three is the "gamut" mapping,
point 2 was impossible to achieve until the ICC decided to specify a black
point in the PCS and point 1 is finally addressed with the introduction of
the adaptation tag for white point. It's coming up on 10 years and the ICC
is finally addressing issues that have been well understood in the
photographic industry since 1952. There is hope.....
Tom Lianza
Technical Director
Sequel Imaging Inc.
25 Nashua Rd.
Londonderry, NH 03053
email@hidden
_______________________________________________
colorsync-users mailing list | email@hidden
Help/Unsubscribe/Archives:
http://www.lists.apple.com/mailman/listinfo/colorsync-users
Do not post admin requests to the list. They will be ignored.