🚀 go-pugleaf

In article <1993Jan4.134820.1522@tom.rz.uni-passau.de> friedric@diogenes.uni-passau.de (Carsten Friedrich) writes:
>I have a big problem converting colors from cmy to rgb and vice versa.
>The usual aproach, to just invert the components seems to be incorrect.

I think it's more a problem of media than the algorithm.

There's a lengthy discussion of color theory in Foley, van Dam,
Feiner, and Hughes, _Computer Graphics: Principles and Practice,
Second Edition_.  You might also want to look at a recent (within the
last couple of years) issue of the ACM publication Transactions of
Graphics.  There was an excellent treatment of the whole subject of
differences between kinds of media and their effects on color.
(Sorry, my copy seems to have evaporated from my shelf, so I can't
give an exact reference -- the cover had a computer generated wine
glass on it, though.  Perhaps someone with better control over their
library can fill you in.)

Finally, if you have a Macintosh, you might try fishing up the San
Diego Supercomputer Center's color theory tutorial software.  Holliday
Horton did an excellent job on it.  It lives at the sdsc.edu
(132.249.20.22) anonymous FTP site in
PUB$:[ANONYMOUS.SDSCPUB.APPLEMAC.GRAPHICS] (yeah, sdsc.edu is a VMS
VAX), and is called INTERACTIVE_COLOR.  It is in both StuffIt and
BinHexed StuffIt formats.

This should keep you amused for a while...

							spl
-- 
Steve Lamont, SciViGuy -- (619) 534-7968 -- spl@szechuan.ucsd.edu
UCSD Microscopy and Imaging Resource/UCSD Med School/La Jolla, CA 92093-0608
"Living alone is living in perpetual irony and thinking, more and more
frequently, Why not."   - Joyce Carol Oates, "The Boyfriend"

In article 1522@tom.rz.uni-passau.de, friedric@diogenes.uni-passau.de (Carsten Friedrich) writes:

>
>I have a big problem converting colors from cmy to rgb and vice versa.
>The usual aproach, to just invert the components seems to be incorrect.
>
(...)
>
>carsten

Did you try this:

	R = V/cv + Y
	B = U/cu + Y
	G = (Y - 0.299 R - 0.114 B) / 0.587

Probably you forgot these constants cv = 0.493 and cu = 0.877.

Reference:

"Fernsehtechnik; Bd. 2: Systeme und Komponenten zur Farbbilduebertragung",
 Broder Wendland, Hartmut Schroeder, Huethig Buch Verlag Heidelberg.
 pages 51-53.

Carsten Salewski

email: salewski@helios.tnt.uni-hannover.de

In article <1993Jan4.134820.1522@tom.rz.uni-passau.de> friedric@diogenes.uni-passau.de (Carsten Friedrich) writes:
>Hi,
>
>I have a big problem converting colors from cmy to rgb and vice versa.
>The usual aproach, to just invert the components seems to be incorrect.
>The following facts make me think so:

>- if you print a bar of yellow on the screen with brightness ranging from 
>0% to 100% you'll see that it is slightly greenish

You are using (r,g,b) = (1,1,0) of the CRT ( using its own phosphors )
to "define" yellow. This has nothing much to do with any other definition of
yellow. You are working in device dependent space and you have to measure your
CRT to derive the device independent data in your favorite device-independent
color space.

>- if you look at an image that is scanned with a very good scanner 
>(like hell) you will see that e.g. red colors in real turn to purple colors 
>on the screen

Do you mean that colors you perceive to be red ( under what illumination )
on film/paper appear to have changed hue on the CRT? Even for device independent
descriptions, cross-media matches are very difficult and a field of ongoing
research. Means: don't expect a match.

>- if you look at color tables printed in 7 color-printing technique you'll see
>that the inverse color of cyan is not red, but a kind of orange. Also the inverse
>colors of magenta and yellow, are not the green and blue you use in rgb.
>if you trust the color tables it seems not to be just a linear offset, but sth.
>more complex. 

Your basic problem seems to be that you are operating in device dependent
color spaces ( rgb and cmy or even better cmyk are generally created with a
specific output medium/dye-set etc. in mind).

>Can anyone help me solve this problem ???

Even a rough color calibration to any device independent color space should
reduce the problem dramatically. You might try a device independent rgb ( like
Xerox rgb 8-) ) which does in general not span the available gamut, or a
luminance chrominance system like YCC,YES,YHS,Y??.

>thanx and greetings,

>carsten

Reiner