Re: Weird Color Behavior?
Re: Weird Color Behavior?
- Subject: Re: Weird Color Behavior?
- From: bruce fraser <email@hidden>
- Date: Sun, 22 Feb 2004 12:39:54 -0800
At 11:38 PM -0500 2/21/04, Roger Breton wrote:
Of course it's not very useful in itself but it's fantastic to illustrate
RGB working spaces!
RGB working spaces are theoretical constructs. When we look at white,
we adapt to it, whether it's a 5000K white, a D50 white, a 6500K
white, or whatever. The Exploratorium in San Francisco has a nice
exhibit that illustrates this fairly dramatically. It pops up a white
card (or so it seems to the eye). Then it pops up another card that's
whiter than the first one, and suddenly the new card looks white and
the first one looks warm gray. Then it pops another, which makes the
second card appear cool gray and the first card a darker warm gray.
Then it takes the second and third cards away, and the first card
looks white again. Of course, instruments can't do this kind of
adaptation, so they can detect the difference between 5000K and
6500K, but our eyes don't see that difference unless we set up
artifical situations to make them see it.
Now, if you tell me that's how blue an sRGB document
would be displayed on my D50 calibrated monitor with an option to display
the document white absolutely matched to my monitor then I'd say no thank
you, of course.
That is indeed how blue it would appear. The effect is perceptually
exaggerated because
a) unless you hide all the white UI elements on the display, your eye
is still adapted to 5000K white, and
b) monitors can only change the color of white by displaying
something less than their full brightness, which further exaggerates
the color.
And I have to confess that this partly does what I'm after,
thank you. The way Photoshop normally display an RGB document now is that it
manages all the document colors except the document white.
No, it manages all the colors INCLUDING document white. It scales
source white to destination white, then maps all the other colors in
relation to that destination white, because that destination white is
what your eye will adapt to when you look at the display.
But the way an
AbsCol conversion from AdobeRGB to my D50 monitor profile comes out is that
all colors are now exagerately blue. I'll do some measurements and compare
with what I expect those colors to display.
Unless you have a 2nm spectrodariometer lying around, you're gonna
get very noisy data. But I think you're also mixing up the roles of
horse and cart. All the measurement-color-geek-analysis activities we
go through are aimed at producing the correct visual appearance. You
already know that absolute colorimetric conversion isn't going to
give you that.
The object of the exercise is to get the working space to display as
close as possible to identically on everyone's monitors.
Hmmh?
The whole point of introducing working spaces and transforming the
display signal to the local monitor profile is to remove the
discrepancies between people's monitors. Back in the dark ages of
Photoshop 4 and beyond, Photoshop just sent RGB straight to the
display. We tried to get all displays to produce the same appearance
by calibrating them all to the same standard and, hey, it didn't
work. Relative conversions to your specific monitor do work when your
eye is adapted to that monitor white and you have a profile that
describes the monitor's behavior accurately.
I agree that's the intent. Now, whether this actually 'lets us all see the
same thing' irrespective of our preference for monitor calibration I am not
sure? My humble experience suggests otherwise. The example I have in mind if
the display of a Macbeth ColorChecker II, in Lab, on a D50 and D65
calibrated monitors, side by side. I am sure you've done this experiment
once but would you say that the image appears the same in both cases? Maybe
you will say that we don't all use two monitors we simultaneously view
images side by side, at different calibration, and therefore it's a moot
issue. Maybe I should shut up or stop taking that stuff I've been taking <?>
and simply accept it as a 'fact of digital life', but I have a hard time
with the idea that Photoshop and our eyes are combining into making us 'all
see the same thing' when I can the real difference between different monitor
calibration is plain to see, there is a limit to what our eyes will actually
adapt to, no? In a side by side comparison, can you say that you really see
the 'same thing'?
No, because your eye can't adapt to two different white points
simultaneously. But if you move one of the moniitors so that you can
only look at one at a time, making the comparison a before-and-after
rather than a side-by-side, you'll find that the mismatch pretty much
disappears because your eye very quickly adapts to the differing
white points.
Well, the whole point of my reflexion, and thank you for walking this route
with me, is th whole notion of white. I thought I had a clear mental picture
of how white is modeled and processed in the ICC model between whathever
source and an output profile. But it's how white is handled between the
working space and the monitor that bugs me. I was taking certain things for
granted and all of a sudden I felt I stood on shaky grounds. So that's why I
felt the urge to reach for help. This being said, the point of an imaging
system that displayed white as either 'yellow' or 'blue' would serve the
purpose of developping a sensitivity that all soft proofing is never to be
taken absolutely. Yes, converting relatively to it is probably the only
sensible way to do it but, to me, that ought to be explicit rather than
implicit. Have you never felt that faced with the multiplicity of existing
papers today all exhibiting countless nuances of 'white' that 'white' has
become an oxymoron? What is white: PCS white? It's an ideal that does not
exist. Yet, we all refer to it as 'white', the perfect non-selective
diffuser.
I think the only rational answer to that question is that "white" is
whatever your eye happens to accept as white at this moment. If
you're looking at a print on paper that your eye can reasonably
accept as white (i.e., not goldenrod or salmon), then for you, that
paper white represents the current state of white in your visual
system. That's why when we proof, we do so on a substrate that's
brighter than the eventual destination. It's also why we have to trim
off any uninked areas of the paper to allow our eye to adapt to the
simulated white on the proof. Likewise, it's why we have to hide all
the white UI elements to get Photoshop's "Simulate Paper White" view
to look like something other than a disaster.
--
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.