Re: Chromix ColorThink not using chromatic adaptation tag for plotting monitor profiles (?)
Re: Chromix ColorThink not using chromatic adaptation tag for plotting monitor profiles (?)
- Subject: Re: Chromix ColorThink not using chromatic adaptation tag for plotting monitor profiles (?)
- From: Graeme Gill <email@hidden>
- Date: Wed, 20 Jul 2005 11:53:03 +1000
- Organization: Argyll CMS
Steve Upton wrote:
But, add a display or working space profile with a non D50 white point and it gets murky. My
thinking in the past was to render the colors the same way as those for input and print profiles
using absolute colors. Those of you who use ColorThink know that this results in a gamut volume
that has a non D50 white point. The white point for sRGB, for instance, appears blue-ish and
off-axis. Is this correct? There are certainly arguments that support it. Is it the most
accurate? That's very tough to say. Is it the best for gamut comparison? Perhaps, perhaps not.
In thinking this through I came up with one clear idea. That no matter the monitor white point,
to the "adapted eye" it will be the white reference (white point extremes aside) and all colors
will fall out relative to it. Therefore it could be valid to produce a "relative" gamut volume
for such devices where the white point aligns with the D50 axis. As a result of this rethinking,
future versions of ColorThink will plot monitor / working space volumes as relative colorimetric
colors. Will this affect how the gamuts overlay print gamuts? yes. Is that a good thing? Yes, I
think so. I think it is the best choice to make when plotting multiple gamuts for comparison.
I don't see any problems with displaying a relative gamut comparison, as long as
you don't call it an absolute comparison. An absolute comparison has some interest
in that it shows the total transformation needed between the two devices.
Normally I'd want to be able to see both. I'd certainly want to be able
to see the absolute gamut of a device, not just it's adapted version.
It is interesting to note that Adobe has made a change in Photoshop CS2 that is along the same
lines. That is, if you choose to convert from a non-D50 space like sRGB to a print space using
absolute colorimetric, you no longer get a blue white point as you did in earlier versions but
rather a D50-source white point that is more in line with what people expect. Can it be argued
that the other way is more correct? probably. But the reality is that people didn't get what they
expected AND what they got was not of much value to them.
Sounds dumb. People got what they asked for (when selecting absolute), and
then when they were surprised, Adobe complicate it by making absolute act like
it's relative. So you now have perceptual, saturation, 2 relative intents !
How do you select absolute intent if you need it ?
If the purpose was to introduce additional choices of gamut mapping, then that's
what should be done. Add a new "virtual" intent (say, "absolute except for emissive
displays which will be treated as relative"), don't replace the current real absolute.
The concept of a "true" gamut becomes rather foggy the closer you get to it - rather like trying
to pin down an electron. ColorThink is not currently resurrecting your non-D50 white point, this
is true. But it is likely giving you a better gamut for comparison than had it done so.
I disagree. There is a very clear concept of the gamut of the actual XYZ colors measured
for a device (modulo a Y scaling), and introducing an "absolute" intent that has an
adapted white point of D50 completely throws this away. For the purposes of applying
important transforms (such as Color Appearance Models), it's actually necessary
to be able to recover the real absolute response of the device.
It's not actually the emissive displays that have the "problem". Their absolute
response is very clearly defined. It's actually the reflective media that causes
the problem, since the light source is not part of the media, even though it's
part of the viewing experience. What gets captured in profiling a reflective
media is really it's reflectance values (ie. absolute values divided by
the measuring light source multiplied by a standard - D50). This is a pretty
practical thing to do, given that when you illuminate a reflective media,
the viewing light source effectively gets multiplied by the reflectance.
It's consequence though, is usually to make the absolute reflectance white be
very close to the relative white value, leading people to assume that
the difference between absolute and relative is usually subtle ("just the paper color").
The problem of treating emissive displays absolute intent differently
to reflective media becomes evident if you were to compare a profile
of a D65 display, with the profile of some blue tinted paper.
The "absolute" white point of the display would be regarded as D50 ("white"),
while the absolute white point of the paper would be distinctly blue,
something that in practice the viewers eye would tend to adapt to (in spite
of the light source being D50.)
We have a weekly TV guide here that is printed on green tinted paper
(the "Green Guide"), and it's amazing how natural photo's and such things
look, in spite of the white point being distinctly green. The human
eye+brain compensates.
[The above couple of comments are really about the changes to the ICC spec.
and the practical consequences that have flowed from that, rather than a comment
about ColorThink.]
Graeme Gill.
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Colorsync-users mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden