🚀 go-pugleaf

In article <4566@m1.cs.man.ac.uk> lilleyc@cs.man.ac.uk (Chris Lilley) writes:

   The new red book suggests that colours can be specified in CIE XYZ space
   simply by parameterising the existing CIEBasedABC colour model with the
   white point.  Like this

	...

   Anyhow, this raises several questions which I had hoped someone here would
   know the answer to:

   A) The interpreter must store the CIE chromacity (xy) or tristimulus
   (X,Y,Z) values for its inks internally. Indeed it must also store the
   values for the secondaries too, since the (x,y) point of a secondary is not
   on the line joining the (x,y) points of the two corresponding primaries.
   *So* how do you get the interpreter to divulge these values?

   B) what happens if you specify a colour that is outside the printer gamut?

   C) what happens if you specify a coordinate that isn't even a colour? Like
   X=0.7 Y=0.1 Z=0.2, which gives x=0.7, y=0.1 and is outside the flat edge at
   the bottom of the 'tongue'. You will need to look at a diagram to udrestand
   this, and I'm not going to attempt one in ascii!

   Any answers are welcome, either email or post, and yes I will summarize any
   answers in a later posting.

	   Chris

   PS Manufacturers of colour level2 PS devices are especially welcome to
   reply! :)  Adobe is hereby extended a cordial invitation to provide an
   authoritative answer.

Your are kidding right (:-)?

(a) Most present day color PS interpreters make only a rough approximation
    of the inverting the non-linear relationship between dot-coverage in
    each ink and tristimulus values (or whatever your favourite color
    coordinates are). 

(b) They would never tell you how they do it.  Some companies have made
    their living off knowing how to do this well.  Why would they tell you?

(c) The `science' behind all this has gotten obscured to the point of
    incomprehensibility by practice. Hence the use of such terms as `primary'
    and `secondary' colors, and `color addition' and `color subtraction'.
    This was all rather well understood when it WAS of interest to the
    scientific community (round about Maxwell's time).  Since then it 
    has become obscured by the practitioners who have a need to give
    short rules for explaining how to do things. 
    
(d) Actually, these days, with memory so cheap, one can do a pretty good job
    interpolating from a three-d table constructed using lots of
    experimental samples.

Berthold K.P. Horn
Cambridge, Massachusetts, USA