[comp.graphics] Dithering and Color Literature.

webber@athos.rutgers.edu (Bob Webber) (03/19/90)

In article <7387@celit.fps.com>, billd@fps.com (Bill Davidson) writes:
> some things are just plain hard to find.  For instance, color
> quantization is glazed over very quickly in a couple of texts but for
> the most part you can't find it in the standard introductory texts.
> There is stuff in the journals but that can be quite difficult to find
> for the beginner (or even the old hand ;-).  This particular subject
> has been handled on the net by the FAQ posting (at least the best
> introductory reference to the subject is there).  Another subject like
> this is dithering which is glazed over in most introductory texts but
> is usually covered quite poorly.  The FAQ posting sugests Ulichney
> which is definite overkill (everything you never wanted to know about
> dithering and were sorry you asked). 

Graphics texts (such as Foley & Van Dam's Fundamentals of Interactive
Computer Graphics (Addison-Wesley 1982) or Roger's Procedural Elements
for Computer Graphics (McGraw-Hill 1985)) have reasonable sections on
dithering for people who just have some grey image they want to view
on a monochrome monitor or dot-matrix printer.  If one has access to a
research library, the journal references that accompany these sections
would be the next stop.  Once one gets past that (either with or
without journal refs), Ulichney's Digital Halftoning (MIT Press 1987)
really is the next stop.  While I am not too interested in hexagonal
sampling, the rectangular parts illustrate well the possibilities
(particularly the discussion of blue noise and error diffusion).  Once
one has digested Ulichney, it is time to learn some Signal Processing.
People with Engineering backgrounds come by this naturally, but it is
alien to most people with Computer Science backgrounds.  For the
Computer Scientist, probably the best presentations of this sort of
thing are in the Image Processing/ Computer Vision texts.  Gonzalez
and Wintz's Digital Image Processing (Addison-Wesley 1987) has been
used at a number of Siggraph tutorials and covers this sort of thing
well.  Clarke's Transform Coding of Images (Academic Press 1985) is
also worth reading.  But, as I say, one has to be serious about
dealing with the math if one wants to get further than Ulichney (the
math is referred to in Ulichney, but there are so many experimental
results, that one can sort of glide over it there).  Clarke also
addresses some of the human-related (psychobiology) aspects of this
(after all, the real task is making an image that looks good to a
human).  Probably at this point, it is worthwhile calling up UMI and
getting a few disserations.  Certainly Kajiya's Toward a Mathematical
Theory of Perception (U of Utah 1979) and Dippe's Antialiasing in
Computer Graphics (U of California at Berkeley 1985) are important
here and the references in the back could keep one busy til the next
century.

For color, again the intro texts are fine for someone who just wants
to throw something quickly up on the screen.  After that, there is
Hall's Illumination and Color in Computer Generated Imagery
(Springer-Verlag 1989).  Once past that, one is again in the realm of
signal processing, optics, and psychobiology.  Kueppers' The Basic Law
of Color Theory (Barron's 1982), MacAdam's Color Measurement -- Theme
and Variations (Springer-Verlag 1985), and Nassau's The Physics and
Chemistry of Color -- The Fifteen Causes of Color (John Wiley & Sons
1983) will keep one off the streets and out of mischief, for a while
at least.  Incidently, Ulichney's book has some interesting references
on color digitization (and most of the other references on dithering
also talk a bit about color).  For dissertations, Phong's Illumination
for Computer-Generated Images (U of Utah 1973), Meyer's Color
Calculation for and Perceptual Assessment of Computer Graphics Images
(Cornell 1986) and Kim's Digital Color Image Enhancement Based on
Luminance & Saturation (U of Arizona 1987) are important and have
good pointers into the literature.

If the destination of the image is a monitor, then Conrac Corp's
Raster Graphics Handbook (Van Nostrand Reinhold 1985) is useful.
Storer, Starr, and Valley's Cathode Ray Tube Displays (McGraw-Hill
1946) is neat too (indeed, it has the distinction of being the most
modern technical work that is useful and not covered by copyright that
I know of -- the preface says the publishers agreed to release it into
the public domain 10 years after publication).  I haven't heard of
anything comparable to either of these for film or printer
technologies although Phong talks a bit about the problems of getting
the same image on film that one started with on the monitor and
Durrett's Color and the Computer (Academic Press 1987) contains an
article on printers with a few refs in it (the rest of the book is CRT
oriented).  Film technology is mature enough that somewhere I suspect
there is a bible on film response to color, but printer technology
seems to be a bit in flux now (although there are doubtless texts on
the technology traditionally used in the book/magazine industry -- its
just that I suspect these matters are still proprietary for color
laser printers, cycolor, or thermal printers and such).  Kueppers'
talks a bit about the various technologies in general as does Nassau's.

--- BOB (webber@athos.rutgers.edu ; rutgers!athos.rutgers.edu!webber)