Re: Call for user feedback
Re: Call for user feedback
- Subject: Re: Call for user feedback
- From: Henrik Holmegaard <email@hidden>
- Date: Tue, 20 Mar 2001 17:11:38 +0100
OK, time the 'RGB working space' confusion which is breaking the workflow.
Subsetting mechanism : The 'ColorSync Profile' (: Monitor Profile)
scrollbox (: popup) in the Monitors control panel shows the user all
class <mntr> profiles in the ColorSync Profiles folder (one nesting
level deep).
Problem : Adobe is classifying synthetic RGB interchange spaces as
type <mntr> monitor spaces. Users are presented with these spaces by
the Mac OS monitor profile selection mechanism.
ICC Specification : The three classes of device profile (: input,
monitor and output) are seen by many more applications than the
non-device profiles (space, link, abstract and named color). Of the
three classes of device profile only the monitor profile is both
source and destination and optionally matrix-based. Of the non-device
classes the <spac> class requires that the profile be LUT-based and
work as both source and destination.
Interpretation / decision strategy : The Photoshop folks want RGB
interchange profiles with a single gamma in all three channels so
that you can do in RGB much of what you can do in LCh. In terms of
the ICC Specification monitor profiles are defined as matrix-based
with LUTs as an option. So the Pshop folks pick <mntr> as the obvious
class for their non-monitor RGB spaces, and they pick a profile in
the device class (: input, monitor and output) in the first place
because those are the three classes most widely recognised by
applications.
Confusion : See 'subsetting mechanism' above and now also that CMYK
profiles are being called 'working spaces', though ironically printer
profiles are legal only if they come with LUTs.
Solution 1 : Adobe obviously doesn't intend it's non-monitor RGB
interchange spaces to be what Adobe says they are, that is, monitor
spaces. Joseph (I think he's away from his mail and can't answer -
maybe Dan or Franz will help out) was saying a while back that
well-designed matrix-based spaces obviate the LUT-based color space
conversion class of profile. Maybe the ICC specification can be
amended to allow both matrix-based and LUT-based profiles as class
<spac>.
Solution 2: If solution (1) is unacceptable maybe the ICC
specification can be amended to allow matrix-based RGB profiles into
a new class of their own which might be called 'Interchange'.
First Apple introduces the VCGT white point and gamma tag in
ColorSync 2.5 without asking any other ICC members, but with the
laudable aim of curbing the Knoll gamma tool.
Then Adobe introduces the 'RGB working space' concept without
squaring with others, but with the laudable aim of moving RGB
workflows from monitor RGB to interchange RGB.
Morale : It's better to make things work across the board than make
them work only in one's own application -:).
--
Henrik Holmegaard
TechWrite, Denmark