[metapost] defaultcolormodel

[Reinhard Kotucha]

On 2011-07-08 at 23:50:26 +0200, Hans Hagen wrote:

 > On 8-7-2011 11:24, Reinhard Kotucha wrote:
 > 
 > > Black, however, is well defined.  It's what you see if you are in
 > > a closed room and turn the light off.  This and the fact that
 > > color pigments used in black ink (inexpensive carbon) are nearly
 > > ideal, make RGB(0,0,0) -> CMYK(0,0,0,1) a very good choice.  You
 > > don't see a difference between CMYK(0,0,0,1) and CMYK(1,1,1,1)
 > > though with the latter you waste a lot of ink.
 > 
 > I'm sure there is a difference between (0,0,0,1) and (1,1,1,1)
 > ... black vs brownish.

Not if black ink is ideal.  And brown shouldn't be CMYK(1,1,1,1), not
even CMYK(1,1,1,0).  The latter might look brownish, but this is
caused by non-ideal inks.  Nothing is perfect, but black ink is nearly
perfect.  If in doubt, ask your chimney sweeper. :)

 > > In LaTeX's color package, black is already CMYK(0,0,0,1), and this is
 > > also a good choice for Metapost.  Please note that with CMYK(1,1,1,1)
 > > you don't not only waste ink, the paper also gets too wet.
 > 
 > GRAY(0) is better, and also less code.
 
Sure, but the question is what to do if no color is specified at all.
In this case TeX and Metapost default to black and it matters how
black is defined in CMYK color models when RGB is converted to CMYK.
In RGB it's easy because all you have to do is to turn all lights off
in order to get black.  In CMYK you have several choices, but given
that black color pigments are ideal, there is only one reasonable
choice.

 > > All devices except printers and maybe nowadays 'electronic paper' are
 > > using an additive color model.  Hence, RGB wouldn't be a bad choice.
 > > But the gammuts are not necessarily the same.  There might be still
 > > differences on what is red, green, or blue.  Sensors might have their
 > > maximum sensitivity at different wavelengths.  You'll see this if you
 > > compare red LEDs from different manufacturers, for instance.
 > 
 > this is when color profiles kick in
 > 
 > > A better approach is to use a standardized color model and put
 > > the calibration data (color profiles) into the devices.  But then
 > > manufacturers have to agree on a standardized color model, which
 > > is unlikely to happen.
 > 
 > there are some standards

... and it would be nice if people make use of them.

Regards,
  Reinhard

-- 
----------------------------------------------------------------------------
Reinhard Kotucha                                      Phone: +49-511-3373112
Marschnerstr. 25
D-30167 Hannover                              mailto:reinhard.kotucha at web.de
----------------------------------------------------------------------------
Microsoft isn't the answer. Microsoft is the question, and the answer is NO.
----------------------------------------------------------------------------