Re: CMYK to Lab
Re: CMYK to Lab
- Subject: Re: CMYK to Lab
- From: Henrik Holmegaard <email@hidden>
- Date: Wed, 4 Apr 2001 18:19:09 +0200
Another and simpler way of looking at the deviceColor / CIEBased /
ICCBased workflow problem is that the creative tools which define
input into the final production process are designed from the ground
up for the studio whose needs to some extent conflict with the needs
of the final production process.
The Illustrator customer is an illustrator (... surprise -:)). And an
illustrator works as follows: First a color draft is created by hand,
then it is approved, and lastly it is transferred to a digital
design. The colors are picked from a printed swatchbook, both CMYK
and spot colors. The CMYK swatchbook may be one printed by the
publisher's printer, or a commercially available Euroscale or SWOP
color bible. The way the Illustrator folks put it a while back, the
Illustrator customer doesn't want the specified numbers to change. In
a device independent conversion you change the numbers to preserve
the colors, but if the illustrator who is at the bottom of the
graphic arts food chain were to do this, he would be responsible for
the colors on the press. So there is no way you will get her to do
that. Hence Illustrator doesn't never supported the in-RIP model and
is only now moving to ICCBased. The RIP doesn't recognize ICCBased
color space specification and treats the objects that reference them
as device colors.
The Photoshop customer is in a different situation. She can't start
with a hand colored drawing, but is forced to start with a digital
painting or a retouched finescan. If you disregard the whole issue of
preserving color appearance, the other face of a device independent
conversion is that you get the right total area coverage and black
generation for the state of the device. I have seen awful CRD
separations from an older PSL2 v2014 printer here in the color lab,
but I'd think the mechanical parameters were probably OK. Back in
1990 there was no way you were going to get a photographer to wrap
her head around Under Color Removal, Gray Component Replacement, dpi
and lpi, graybalance and goodness knows what - in addition to her
real job of figuring out F-stops, film color bias, chatting up
clients and so forth. So the Photoshop CSA to CRD implementation did
what made sound horse sense many years ago by automating the
mechanical separations for people who weren't in the business of
building separation curves.
Part of the solution is as Jan-Peter Homann suggested that Adobe
prevents CIEBased spaces from being embedded by Photoshop.
But there must be many many users in networked environments where
hotfolders are used who generate PDF by printing PostScript from
Illustrator 9, InDesign 1.5 and Photoshop 5.0, 5.02, 5.5 and 6 into a
Distiller hotfolder. If they do this and select PostScript Color
Management in the printing dialog along with a rendering intent to
strip out of the ICC profile and convert into a CSA, then the Adobe
application converts the source ICC profile (AI9 in RGB mode a
matrix-based RGB working space available as an ICC profile on the
disk, in CMYK mode the LUT-based ICC printer profile shown in the
CMYK Working Space and made by e.g. Heidelberg Printopen or
GretagMacbeth ProfileMaker), and turns that into a CIEBasedABC and
CIEBasedDEFG, respectively.
Distiller 4.05 will preserve the CIEBasedABC in PDF 1.3 and convert
the CIEBasedDEFG into Lab. (Distiller 4.0 broke ICC printer profiles
when embedding them, the profile was there but mangled. It's also
confusing for users that Distiller can't read incoming ICC profiles
but can embed them into the outgoing data stream. Adobe calls both
ICC profiles and CSAs 'profiles', and that doesn't help one tiny bit
either.)
This is all very easy to check with Adobe InProduction 1.0, even if
InProduction can do nothing about it. The reason is that InProduction
only lets you tag / untag / convert for device dependent color
spaces, but it can't select Lab objects and assign an action to them
alone. For this you need GretagMacbeth iQueue 140 which like the
older BatchMatcher tool has the ability to assign ICC actions to Lab
alone, CMYK alone, RGB alone, Gray alone, or any combination.
When you try to create a bug matrix for ways to produce a PDF 1.3
that will work as intended, your time is better spent going fishing
-:).
--
Henrik Holmegaard
TechWrite, Denmark