🚀 go-pugleaf

In article <4591@optilink.UUCP> cramer@optilink.UUCP (Clayton Cramer) writes:
>In article <339@aupair.cs.athabascau.ca>, kevinc@cs.athabascau.ca (Kevin Crocker) writes:
>> In article <245@srchtec.UUCP> mra@srchtec.UUCP (Michael Almond) writes:
>> >>> According to PC Week, 32,000 colors. They also said it's 16-bit
>> >>> color. I don't really understand this.
>> 
>> If we are talking about ~32000 hues on screen then is there any
>> evidence to show that the human eye can discern this level of sublety?
>> 
>> Kevin "auric" Crocker Athabasca University 
>
>I read somewhere, many years ago, where I can doubtless not find
>it, that the human eye is capable of distinguishing about 10 million
>shades of color.  (Presumably measured by taking two light sources
>of known frequency some number of Angstroms apart, and seeing how
>many people can distinguish them).
>
>This means that when you get to 24 bits of color information, there
>is no point in going any further.
>
>-- 
>Clayton E. Cramer {pyramid,pixar,tekbspa}!optilink!cramer
>If "right of the people" in the Second Amendment doesn't refer to an
>individual right, what does it refer to in the First & Fourth Amendments?
>You must be kidding!  No company would hold opinions like mine!

Actually, this is not true.

We're talking my field here; I design color scanners, color correction
systems, and work on the monitor to print problem, for Xerox.

24 bits is insufficient for more than one reason.

The first thing to consider is that the dynamic range of the eye is not
a simple function, and is not the same for each of the three (actually,
four; did you know that under the right circumstances people can be
tetrachromats? This is called *rod intrusion*) input channels (which
may be equated with red, green, and blue sensation).

Under good conditions, with a calibrated monitor, you can see banding
in an 8 bit gray ramp.

Also, the range of intensity that is expressed in that ramp does not cover
the full range of intensity the eye can percieve; i.e., there are
"whiter whites" that you can respond to.

The situation gets even more fun with the addition of colors, as
opposed to grays.

Another fun complication is that the primaries used in video, print, etc.,
do not cover the full *gamut* of color vision. If you plot the gamut
given by, say, the SMPTE phosphours on a CIE color chart you will get a 
triangle inside the full range of human vision. There are colors you can
see perfectly well that no monitor can display. This gives rise to a whole
host of interesting tradeoffs - do you clip colors to the hull of the gamut,
or somehow map one gamut to another? We do *both* here, depending on the
image and the device - for example, last year's SIGGRAPH proceedings
cover had a gamut mapping applied to it from the video image to map it
into the printer's gamut.

Consider also that when you do this that you are squashing levels and
will lose discernable differences. An important consideration from
an artists view is perserving discernable differences even if there
are hue shifts!

...these days I get somewhat exercised by the widespread use of the
phrase "True Color" in reference to 24 bit displays; they are anything
*but* true color. The sad truth is that very few people have much
sophistication for color; it's rather like typography before 1984.



---

Rob Tow
Member Research Staff
Electronic Document Lab
Xerox PARC
3333 Coyote Hill Drive
Palo Alto, CA 94304
(415)-494-4807