RE: A new myth > Rant

["Richard Wolfson" <rwolfson@LyricDesign.com>]

<x-charset iso-8859-1>Robert L Krawitz wrote:

> The issue here appears (to me, at any rate) to be exactly the same as
> RGB vs. CMYK, except one level deeper: CMYK doesn't say anything about
> how C splits up into dark and light cyan.
>
> What it sounds like your point is, though, is that the profile is
> essentially a mapping between input CMYK and printer output; if the
> output is non-linear, a good (i. e. complete) enough profile will
> correct this through feedback (in the profiling process).

That's the idea, exactly -- except it's even better than that. The CMYK
profile (at least the kind I've been working with) is actually a mapping
from the *RGB* image space to the printed CMYK color. Unlike RGB
profiles, which are applied when printing, the CMYK profile is applied
to separate the RGB 3-channel image into a CMYK 4-channel image, which
is then printed "straight through" to the driver without further
application of any profile.

>    Why does the prospect of mixing light cyan and black trouble you?
>
> Partly because it increases the total amount of ink on the paper in a
> range where the paper's already close to saturation.  At that point,
> the inks interact quite strongly with each other, and simply
> increasing the amount of ink doesn't increase the darkness any more.

Point taken. I'd love to know what Steve Upton says about this: he is
far more expert on color gamuts than I. (I've CC'd him on this.) My
hypothesis is that if the image really calls for a very dark color, as
you say, the C value coming to the driver *would* map to dark cyan, not
light, and the partitioning curve need not change shape as a function of
black level. Of course it should phase over from light cyan to dark
(only) above some mid-range C value. I agree that the "adaptive
partitioning" idea is interesting: if there's enough black ink around,
just use dark cyan, since nobody can see the dots anyway, which saves
total ink. I'm just not enough of a color geek to know if it really
helps, and how well it would work with the ICC profile model. I'll stay
tuned for the experts.

Richard Wolfson
rwolfson at LyricDesign.com



-
Turn off HTML mail features. Keep quoted material short. Use accurate
subject lines. http://www.leben.com/lists for list instructions.

</x-charset>