[argyllcms] Re: [EXT] Re: Comparison of Color Profiles to Human Perceptual Gamut
- From: Ben Goren <ben@xxxxxxxxxxxxxxxx>
- To: argyllcms@xxxxxxxxxxxxx
- Date: Wed, 2 Feb 2022 13:17:00 -0700
On Feb 2, 2022, at 12:30 PM, Mizen Mark ("MMizen")
<dmarc-noreply@xxxxxxxxxxxxx> wrote:
[Patrick Herold from Chromix] had a file called “Human percep gamut.txt”
which contained the outer limits of human visual perception,
So … a good way to start to describe “the outer limits of human visual
perception” would be with the octave-ish range of EM radiation we can perceive,
which spans from about 800 nm at the very dim reddish low-energy end to a bit
more violet (and again very dim) than 400 nm at the high-energy end.
Next would be to indicate that we’re most sensitive to green light at roughly
the peak of the Sun’s SPD, in the neighborhood of 535 nm.
From those two points, it’s only a few very reasonable steps to CIE 1931 and
its descendants.
Subjectively, the outer limits are the monochromatic colors of the rainbow. (In
a lab setting; as pretty as the ones in the sky are, they aren’t very
monochromatic.) That gives you basically a two-dimensional plot of the “shadow”
of a 3D volume; the full volume tells you that, at the extremes of brightness,
everything fades smoothly into black and white.
One could then go on to plot individual colors as points in such a volume. And,
further, one could outline the limits of various devices in their ability to
reproduce colors within the volume.
One great tool for doing this sort of thing is the ColorSync Utility bundled
with MacOS. I’m not familiar with Chromix’s offerings, but I’m sure they’ve got
tools that do this very well.
Most high-end printers and displays these days have gamuts that cover anything
you can make with Pantone inks — with the usual caveats about their metallic
and other “special” inks. (They also have some that are highly metameric,
displaying dramatically different colors depending on the spectrum of the light
they’re viewed under.) Out-of-gamut Pantone colors generally aren’t nearly as
much of a concern as they used to be at the dawn of the desktop publishing era
and for quite some time afterwards.
From what I have learned, Pointer’s gamut may well be a reasonable
representation of the human perceptual gamut.
Eh … no. Not even close.
Pointer’s gamut represents the limits of colors one can get from reflective
surfaces when lit by a standard illuminant. Think of it as the ultimate
painter’s palette. But, for example, shine a laser pointer on a white sheet of
paper and you’ll get a color far more saturated than anything in Pointer’s
gamut. Fluorescent dyes lit by UV-containing light will likely be outside of
Pointer’s gamut, too.
I could not find Pointer’s gamut as an ICC profile so I am not sure if it
exists in that form or if there is a reason it does not.
So … ICC profiles don’t do what you seem to think they do.
An ICC profile, essentially, does one of two things. Either it maps from RGB
values captured by some input device (camera, scanner, etc.) to an absolute
color space (such as Lab or XYZ); or it maps from an absolute color space to
the RGB (or CMYK, etc.) values an output device should use to best represent
that color.
(That’s gross oversimplification, of course, but “good enough” for this
discussion.)
So there’s no way to represent Pointer’s Gamut as an ICC profile. It’s neither
an input device nor an output device.
Similarly, there’s no way to represent the limits of human vision with an ICC
profile, though one could make an argument that any absolute color space (Lab,
XYZ, etc.) is itself such a representation. But be careful: for example, XYZ is
fully capable of encoding “impossible” colors that don’t correspond with
anything any human has ever perceived.
Hope this helps …
b&
Other related posts: