 |
|
The Inception of Computer Graphics at the
University of Utah 1960s - 1970s
|
Tuesday, September 27, 1994
Notes by Gianna
Walker
The late 60's through the 70's at the University of Utah was the
golden era of computer graphics research. At that time they had
the luck, money, people (creative grad students and supportive instructors),
and the opportunity to do basic research in the field of computer
graphics. They knew that they were "onto something big"
while outsiders at other universities disparaged the work in computer
graphics as an illegitimate application of computing machinery.
Computing research at that time involved computer languages, operating
systems, and data processing. Graphics research required manipulating
so much data to display images, that it pushed the envelope in computing
technology. Students came from diverse backgrounds, and for them
the discovery was a labor of love. There were no Computer Science
or Computer Graphics fields, nor jobs waiting for them upon graduation.
These students created the computer graphics companies (LucasFilm
computer division, Pixar, SGI, Adobe, etc.) that lead the industry
today.
John Warnock (grad student '68 - '75):
Dave Evans founded the Computer Science dept. at the University
of Utah in 1966 with ARPA funding to study man-machine interaction
($5mil /yr for 3 years). He recruited Ivan Sutherland and other
notable professors. Others such as Alan Kay, Jim Clarke, William
Newman, and Jim Blinn attended UUtah and as alumni went on to work
at Xerox PARC. The work done at Utah was pure research, as classroom
curriculum in the field was almost non-existent. Grad students with
a spark for imagination and creation, come in to try to do "neat
stuff" as their first and only priority. There was a peer-to-peer
relationship between students and professors, often former grad
students. CG models created lasted a long time e.g., the teapot
sitting on Martin Newell's desk. Modelling was not the focus of
study, but the images created with them - "making pictures".
Early input of programs was a key-punch-batch process run on a UNIVAC
1108 that ran twice a day. Then Fortran programs were typed on teletype-like
input machines which created a terrific racket in a room filled
with grad students, yet this was an improvement over the punch card
system. Triple exposed film with R, G, and B filters created color
hardcopy output. Modeling was analytic, not aesthetic. They programmed
parametric equations to build polygons.
Ed Catmull (began '63 undergrad):
As an undergrad took classes from Alan Kay, who was frequently late
to morning class, but exhibited the enthusiasm for computer graphics
which inspired the students. At that time so little information
was known, that often speculative (wrong) information was given
as fact during a lecture, only to have it overturned by students
in lab. Ivan Sutherland gave one class a project to produce a movie
by using an existing program or writing one from scratch. Those
who produced them from scratch, went on to make other significant
discoveries. In '73 Ed Catmull digitized a casting of his hand and
animated the 3D computer model (Catmull cast his hand in plaster,
and discovered to his dismay that pulling the cast off also pulled
out all the hairs on the back of his hand). Henri Gauraud developed
"smooth shading" algorithm by averaging the intersections
and vertices of faceted objects to yield a smooth surface, while
others were trying to make the facets more fine-grained. The Gauraud-shaded
images took much less computational power than resolving the number
of facets down to infinity. Yet, while the images were not faceted,
they still "looked like felt". This flat look was solved
by creating a reflection of the light source itself in the 3D object
(specular highlight) with its position determined by the viewing
angle (Phong shading). With this discovery, images began to appear
more photorealistic. Developments like this were helped along at
Utah because the study of image processing and understanding was
done along side computer graphics. Elsewhere, image-processing was
a separate discipline and not integrated with procedural modeling
as it was at UUtah.
Frank Crow: grad student '70s
At that time, the output of images was to printers (showed chess
pieces printed in typewritten characters). J. Warnock created the
1st color display picture (described above). [color images were
shown that used Phong smoothing and specularity algorithm). Hardware
used then, PDP-10 w/ 96K memory timeshared by 30-40 people. The
Watkins Box, hardware for real-time shaded picture systems. Equipment
came to UUtah with Ivan Sutherland from Harvard, funded originally
by the CIA. One setup was an early VR-like prototype: virtual image
on a head-mounted display visor with 20,000V at each temple(!).
Position is tracked with a hand-held wand connected to ceiling wires
to divine position coordinates. Some kind of large 3-space microphone
was shown, but worked only once for 10-min. Finally a frame-buffer
was created to enable the development of paint programs. The teapot
on Martin Newell's desk was digitized and animated. Jim Blinn developed
bump-mapping, and all of this enabled the development of flying
logos for broadcast (early picture of an RGB "ABC" logo
shown).
Lance Williams: ('72 - '76)
Undergrad English-Asian studies double major who came to a UUtah
summer seminar in Humanistic Computation. Jef Raskin was the instructor,
teaching the "flow" language of programming for novices
much resembling a flow chart's structure. Lance Williams created
animated movies and synthesized images, studying light and its interaction
with material surfaces. He showed a video Jules Bloomenthal's animation
of growing trees and rustling leaves using texture-mapped bark and
leaves placed on various planes, bendable by invisible wind.
He read from a list of important areas of inquiry at UUtah circa
1975:
- The unreality of stop-motion photography (as seen in model
animation like Ray Harryhausen's dinosaurs) carried over to
CG. Each exposed shot, though eventually animated, was a still
image - not moving so a series of still shots looked fake and
jerky. This lead to the discovery of "motion-blur"
in computer graphics to simulate the blurred image in the exposed
frame.
- Texture mapping: proposed by Ed Catmull as a way of adding
photorealistic detail to 3D modeled geometry (initially pronounced
"useless" by Ivan Sutherland).
- Digitization of stationary objects and to record motion of
the object's path in space. The correlation of frames in stereo
pairs enabled the user to interpolated the z from the x and
y coordinates thus producing a depth map on a frame-by-frame
basis.
- Interpolating walk cycles, tracking the user's head and displaying
on-screen, mentioned someone's project: "Stereo for one-eyed
guys".
- An "automatic rotoscope" was proposed vs. traditional
Disney animation frame by frame. These sequences shown in class
identified some shortcuts such as character can slide on a sidewalk
in a cycle as long s no high-contrast lines present for the
eye to register to.
- Automatic in-betweener: a vision approach: transitions not
faded or jumpcut, correspondences in contour lines of 2 objects
are interpolated ( = morphing).
- Images generated by soundtracks and vice versa for use in
lip-sync, playing musical instruments, collision detection of
on-screen objects.
Lance believed (in '75) "the prompt featuring of digital imagery
in motion pictures" but needed to find some really rich sucker
to spend the money - everyone laughed.
Funding waned in the late '70's. Defense dollars then must be used
for something defense-related. The ARPANET was an unsecured network
for Universities to collaborate. These 25 "centers of excellence"
received $$$ funding, and attained critical mass which boosted the
US ahead of everyone. The military funding in this era built the
scientific infrastructure to allow the US economy to advance. It
would be difficult to do today since military spending is down and
more strict accountability is taken. The funding was put to use
in creative research unencumbered by reporting and direct military
application. About the reduction in funding for CG research, Lance
quipped, they had come up with "no fatal cartoon" yet.
Although the military's long-term goal of real-time flight simulation
(vs. other funded areas such as AI) did come out of this initial
CG research.
The panelists allowed that the early images created at Utah were
not aesthetic works of art, having been derived from procedural
models. But they are optimistic for future use of CG in the visual
arts due to the improved user interface to enable creative people
more self-expression.
Back to Silicon Valley SIGGRAPH Home page
Send comments to the webmaster
|
|