Date: Thu, 15 Dec 1994 14:45:29 GMT

Archive-name: comp.viz.faq

	"Feed me."  -- Audrey II, the Plant in "The Little Shop of Horrors."

		Contents
		========
		Intro
		-----
		What is visualization?
		----------------------
================================

The FAQ is dedicated to Harold "Doc" Edgerton (MIT).

Intro
-----

The quality of this FAQ is directly proportional to the information
mailed by participants (that's you) to the maintainer (me, Amelia).
You don't like it?  Blame the rest of the posters and readers.
I don't have 100% time to maintain this.  We are starting from scratch.
If an address or phone is out of date?  Blame the group.  I'm just a dumb
computer posting as I am told.

	[It's getting better, people are emailing contributions!]

The structure of this FAQ is currently a twice monthly post, with two
weekly outriggers pointing to the Long (this) version.  We will see how well
this works (empirical science).  The header, the body, etc. have all been
specially designed and tested.  In time, mitosis will occur on this file
and it will get spread over the course of a month.

You should not have to see this file all the time.  Don't bother saving it
unless you have a flakey net connection.   Grab the most recent revision
off the net.  The Subject: line is designed to fit within the 24-char limit
of most Killfile systems.  Learn about Killfiles.  If you have something
against FAQ files, the regular expression /.*FAQ$/ will Kill All My Children
and me.  Learn how to use news and how news works before complaining.
This post is like a lighthouse or a fog horn.  Learn how to use it.



What is visualization?
======================

"Visualization is the use of computer-generated media based
on data in the service of human insight/learning."

Analytic graphics
		--Carol Hunter, LLNL

Visualization: 
	The use of computer imagery to gain 
	insight into complex phenomena.

The purpose of visualization is insight, not virtual realities or pictures.
		--Eugene N. Miya, President, Bay Area ACM/SIGGRAPH,
		  Usenet post, November 1987, responding to the release
		  of the Report.  "I like the Livermore perspective."

The Maintainer gets a chance to pontificate here: [You can, too. Just ask.]

I learned from Jack Estes (UCSB) that observation has three recursive
phases and corresponding mathematical models:
	Detection
	Identification
	Measurement and Analysis (this latter being the recursion).
Any good system will support these (reducibly).  Researchers need the latter.
Artistes only use the first (Calder was an exception, but then he trained
as an aerodynamicist 8^).  We don't teach observation very well in schools
these days.  We assume you have picked it up via osmosis or are naturally
gifted.  It requires training and some talent.  It's especially bad in
computer science depts. because of the emphasis in theory.

It is Miya's assertation that ultimately visualization will converge
of technologies like those used in cartography (>2000 years old),
surveying, photogrammetry (aerial or terrestial), certain parts of ergonomics,
etc.  Why?  Because they are quantitative.  They use stereo in some instances
(far more important than most people realize).  USGS tours are available.

Maps have better information content than any image pervading our society.
The signal to noise ratio is greater (density).  Better than photos,
better than movies (ask yourself if you can always figure out the ending 8^).
I defy the presentation of an other consistent image with better
characteristics than a map.  Maps have excellent characteristics like
an easily measurable geometry, legends, symbolization (but like all imagery
they have limitations).  Consider all this the next time you look at your AAA
maps when you go on vacation.

What kinds of quantitative, numeric support are needed?  Every basic science
measurement:
enumeration (counting, math got started because of counting), distances,
areas, volumes, angles (planar and solid), extrema,
parametric and non-parametric statistics (means, median, modes,
deviations, ANOVA), histograms, intervals and error bounds, derivatives,
partials, integrals, etc.

ONLY the trained researcher (maybe you) will make the critical insights
needed for scientific discovery.  Not your programmer, not your 
Renaissance Team.
We are not talking EdVis or PresentationVis, we are talking
HackerVis/NerdVis/DeepDownAndDirtyViz.  We can't do it for you.
You the researcher have the eyes.  Only you can make the discovery:
to see differences where your artist, your programmer, your cognition expert
can't, and to see similarity where the average man can't.

Miya's Suggestion: Never view an image or film, too fast.  You should control
the rate at which you observe.  Use tools if you have to: hand lenses, filters,
etc.  If you are forced to view something too fast, jump up and say:
	"Wait, just one damn minute!"
Bob Sharp (Emeritus, Caltech Div. of Geog. and Planetary Sciences) presents
an Austin Post photo in one book and asserts that this single photo (of Denali)
is adequate for a 1 hour lecture on glaciology.  That's quite a claim (an
impressive photo).  Artificial data images have yet to assert that kind of
information density (it's not clear they should, but these are opinions).
This should be considered an implicit challenge for visualizations to get
that important for discovery.  Sometimes, some fields pour days, weeks, and
months trying to understand (interpret) an image.

Miya's Gross Generalization: Scientists like grided graph paper (and tabular 
papers) like many artists (painters) like starting with a blank canvas
(exceptions always locatable).
Frames of reference are needed badly.  Include everything a good map has:
1) A Scale, 2) a legend (both geometric and any symbology like color or
glyps or icons, whatever).  Beware of perspective.  $Billion$ are spent
yearly to remove perspective from photographic imagery.

Miya's Guess: the biologists will figure this all out before the physicists,
chemists, and others.  Others should attempt to prove this wrong.
Good luck guys. 8^)

Other inspirations: Muybridge and Edgerton.
		Add your own.
	Muybridge did his basic research at Stanford and Penn State.
	His work went on to become the motion picture industry, ergonomics,
	system analysis (time motion studies).  Edgerton is best known
	for his work on the strobe, super-fast and time lapse photography,
	and side-scanning sonar.

Hints on Design from Don Norman  -- used with permission
("Turn Signals are the Facial Expressions of Automobiles")

	A challenge to the designers of the world:
		Make signs unnecessary.
Think of the Four-Questions test and generalize to systems.
Norman originally wrote this test for kitchen appliances, but it generalizes
well to scientific computer systems
	Ouestion one: Where would we store it?
			[Consider disk space as well as foot print.]

	Ouestion two: Where would we use it?
			[Left to you.]

	Ouestion three: Where would we plug it in?
			[Consider more ways that power.]

	Ouestion four: How much work would it be to clean?
			[A question of maintenance.]

	Generalize additional important questions:

Robert Lucky notes the qualities of pictures in Silicon Dreams (pp. 292):

	describing spatial relationships
	showing the structure of data
	allowing pattern matching approaches to problem solving
	getting attention
	describing and identifying people
	invoking esthetic appreciation

This boils down largely to "geometry."

The problems:
	Hidden object elimination (obscured features)
	Optical illusions (leading to false interpretations, e.g. Necker cubes)
	Inadvertant data corruption
	Parallax and perspective in 3-D and higher D.
	Performance

Science differs from art in that we have to validate our simulations and
theory.  Right now, support for this validation is especially lacking.
Empirical/experimental techniques make simulation validation difficult,
in the works of Eduard Imhof speaking about maps:

Chap. 16 page 359.
7. On art in cartography

 The means of cartographical expression are subject to the same experiences
and visual aesthetic rules as every other type of graphic product.  *Art*
however, is the highest level attainable in graphical work.  Thus, a good map
cannot lack an artistic touch.
  There has already been much debate and writing on the question of whether
cartography has anything to do with art and if so, how much.  We must try to
remain in the clear on this topic and avoid exaggeration and cliche.  Certainly 
it is not a function of cartography to create art in the higher sense of the
word: the cartographer has scarcely the opportunity of doing so.  Art
presupposed the widest ranging freedom of form and structure, whereas
cartographers are confined to the smallest details by topographical survey,
statistical figures, by standardization of symbolism and color, and by what
is essentially a non-artistic purpose.  On the other hand, however, the
following facts are clearly established; we demand of it a balanced
expression which emphasizes the significant and subdues the insignificant;
amd we demand a well balance, harmonious interplay of all elements contained.
It is in accordance with practical experience, however, which the author has
personally observed over many decades, that in cartographic affairs, as in
all graphic work, the greatest clarity, the greatest power of expression,
balance and simplicity are concurrent with beauty.  To create beauty, a
purely technical, practical arrangement of things is not sufficient.  Beauty
is, to a large extent, irrational.  Artistic talent, aesthetic sensitivity,
sense of proportion, of harmony, of form and color, and of graphical
interplay are indispensible to the creation of a beautiful map and thus to a
clear expressive map.