Re: Microsoft's color-management claims
Re: Microsoft's color-management claims
- Subject: Re: Microsoft's color-management claims
- From: Graeme Gill <email@hidden>
- Date: Fri, 23 Sep 2005 13:57:02 +1000
Tim Vitale wrote:
If the XYZ PCS uses matrix-based profiles, and WCS will be using XYZ
PCS, then profiles in WCS will be matrix-based rather than the
predominately LUT-based device profiles in the ICC system. Of course,
this generalization excludes the matrix-based ICC working space profiles.
There's no possible justification for drawing such a conclusion. ICC
has XYZ based matrix and LUT profiles, but this says nothing about WCS.
The current information about WCS is high level, and I would suspect
that it talks about XYZ to underline the point that the profiles
are CIE but not white point adapted. The underlying input to any sort of
CAM (like CIECAM02) is also going to be XYZ, since this is the
fundamental CIE colorspace. But it's easy to convert amongst
all the CIE derived colorspaces, and the WCS profile format promises
to be super flexible, so there is every reason to believe that
WCS profiles will use whatever colorspace they like internally (including
CIELab), with conversion to whatever space WCS requires happening
at the end (which may have to be XYZ, or it may not).
The earliest version of ColorSync was based on color equations, XYZ
matrix equations, <http://en.wikipedia.org/wiki/ColorSync> rather than
LUTs. However, as the story goes, it was converted to a LUT-based
system to allow peripheral manufacturers to differentiate their product
lines.
Not a very likely story. Matrix based profiles only work on devices
that have additive behaviour (essentially the colorants don't interfere
with each other). Practically no subtractive devices (such as printers)
behave this way, so matrix profiles work really badly for them. LUT
based profiles on the other hand, are very general, and will work OK
to very well on almost any sort of device.
Assuming the above, it would seem that WCS is closer to the original
intent of ColorSync. It might restore the purity of the original
equation-based color correction system from the current LUT-based
profiles with their "secret sauce" and Lab PCS. OK... I do have problem
with the Lab PCS, so I'm not troubled by trying something new. ...but
when did MS make an improvement in anything...
This is totally off the mark. Matrix profiles may seem "pure" because they
are simple, but that doesn't mean they are good at representing the often
complex behaviour of real world devices.
The problem I have with the Lab PCS is that it makes the following
irreversible changes in the actual color numbers, not the perception of
the color:
(1) desaturates yellows
(2) desaturates greens
(3) some blues go purple
(4) some reds go orange
due to the non-uniformity of the Lab space shown in
<http://www.brucelindbloom.com/index.html?MunsellCalculator.html> and
<http://www.brucelindbloom.com/index.html?MunsellCalculator.html>. Most
folk don't notice these issues, but for me, I spend considerable time
correcting their effects in the proofs, after those great Atkinson
profiles makes the grays neutral on the Epson 9600.
You're wrong to think that Lab space "does this" to the color. It's
just a CIE derived colorspace. If all you do is convert to Lab and then
to something else, nothing happens to your color. If you decide
to do gamut mapping in Lab space on the other hand, then you
might well see some of the above effects. Note that this has little
to do with what colorspace the device profiles are encoded in, since
this is mainly just storage. The colorspace used to do gamut mapping in,
need have no connection with what space the profiles are encoded in.
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