syracuse robin forrest october 2015 comments which fleshed out the presentation have been added to...
TRANSCRIPT
Syracuse
Robin Forrest
October 2015Comments which fleshed out the presentation have been added to the slides
All Thatand
Rich Riesenfeldfrom Graduate Student to Utah Professor
Some history
The early history of CAD is largely
undocumented: there were no journals,
research was reported in Technical Memos,
and a great deal of work was carried out by
industry with little being revealed to an
external audience. Defense funding also played an important rôle.
M.I.T.
Sketchpad, Ivan Sutherland
with Elaine Cohen at the 1984 Steven Anson Coons Award presentation.
M.I.T.
TX-2 and display, Lincoln Labs.
Tim Johnson demonstrating Sketchpad-3
M.I.T.
Computer Aided Design Group, 1960’s
Ross, Coons and Ward
funded by the US Air Force
Doug Ross was the developer of the APT part-programming language
M.I.T.
Computer Aided Design Group, 1960’s
Ross, Coons and Ward
Steve Coons
M.I.T.
Project Mac, 545 Technology Square
the building on the right, off to the east of the MIT Campus, housed IBM, the CIA, the AI Lab and the AED Group as well as the Project MAC CTSS machines
MAC stands for Multi-Access Computer or Machine-Aided Cognition
M.I.T.
IBM 7094 running CTSS
Kludge display
Charles Lang operates the kludge satellite display and Doug Ross explains
M.I.T.
Kludge display
peripherals include joystick, light pen and button box
M.I.T.
AED (Algol Extended for Design
or
Augmented Engineering Design)
Doug Ross
M.I.T. 1964
Ross: AED and data structures
Lang: 7094-Kludge link
Coons: curves and surfaces
in the 1960’s computer graphics researchers wrote papers on data structures as there was a need for more complex structures than arrays, lists and trees
M.I.T.
Charles Lang
Cambridge 1965
Computer-Aided Design Group,
Engineering Department (Welbourn)
and Mathematical Laboratory (Wilkes)
Welbourn had heard about Sketchpad and Wilkes had visited MIT and seen both CTSS and Sketchpad. They decided to form a CAD Group, 50 years ago.
Cambridge 1965
Titan (Atlas-2), time-sharing operating system
DEC PDP-7, 8K 18 bit words and 340 display
(£50,000 in 1965)
Additional 32K memory (~64K bytes)
1965 catalog price $100,000
graphics in the mid 60’s was a rich man’s game but Wilkes somehow
had £50,000 to spend at his own discretion
Cambridge 1965
Gray: data structuresLang: Titan-PDP-7 linkForrest: curves and surfaces
the original members of the group, November 1965Lang returned to the UK from MIT
on the screen, designing a bicubic patch in November 1966note MIT design joystick
M.I.T. Summer 1967
Robin Forrest at M.I.T. Project Mac
to work with Coons
(Martin Richards from Cambridge was at Harvard that summer, developing
the systems programming language BCPL which was used by Xerox
PARC for the Alto work stations and inspired Bell Labs to develop first A
then B then C)
M.I.T. Summer 1967
Coons on sabbatical at Harvard
working with Ivan Sutherland
I would commute to Harvard weekly, spending a couple of hours with Steve
and Ivan and then spending the rest of the day with Bob Sproull, Danny
Cohen and Ted Lee
General Motors 1969
Bill Gordon
Mathematics Department
General Motors Research Labs.after completing my Ph.D. thesis in August 1968, I toured
aerospace and automobile companies in the US before and after the March 1969 Computer Graphics Conference at the University of Illinois
General Motors and Bézier 1969
“The crazy way Renault design curves”
I was given a demonstration of the Renault technique: points were dotted around an IBM 2250 display (a clumsy display costing $250,000) and a curve appeared anchored to the first and last points but not interpolating the other points which could be moved to distort the curve - the polygon was not drawn
Bézier had presented a paper earlier in 1969 at a Society of Automotive Engineers meeting in Detroit
Bézier 1969
the following slides illustrate the notation and
formulation as published by Bézier up to the mid 1970’s
V denotes vertices, P denotes points on the curve and
a denotes coefficients.
a0=V0
ai =Vi −Vi−1,1 ≤i ≤nP 0( )=a0 =V0
P 1( )=Vn = aii=0
n
∑
Bézier 1969
P t( ) =a0
+ iai=1
n
∑ nif t( )
Bézier 1969
nif =
−−t( )i
i −1( )!•d i−1Φn t( )dti−1
Bézier 1969
Φ n t( )=1 − 1 − t( )n
t
Bézier 1969
P t( ) = Vii=0
n
∑n
i
⎛
⎝ ⎜ ⎞
⎠ ⎟ ti 1 − t( )n− i
Cambridge and Bézier 1969
I re-cast Bézier’s formulation in terms of polygon vertices rather than polygon sides, revealing he was using
Bernstein polynomialsthis explained the variation-diminishing properties of Bézier’s curves
Summer 1970, Rich sent by Steve Coonsfrom Syracuse University to work withthe Computer-Aided Design Group
Coons moved from MIT to Syracuse around the beginning of 1970, I think
the CAD Group was an eclectic bunch, including visitors from industry - aDutch visitor said we had our heads in the clouds and our feet on theGroundwork in progress included develoment of Bézier’s method and preliminarywork on a B-rep solid modelling system
Cambridge and Rich
Spring Semester Bill Gordon on sabbatical
from GM to work with Coons
Syracuse 1971
Observing that Bernstein polynomials
are a special case of B-splines,
he suggests the use of B-splines for CAD
as a thesis topic for Rich
Syracuse 1971
Fall Semester 1971-2, Robin Forrest
on leave from Cambridge to work with Coons
the original intention was to work with both Coons and Gordon, but Gordon
had decided to return to Detroit earlier than planned
Syracuse 1971-2
Tasked with encouraging Rich
who was under pressure to produce a thesis
Syracuse 1971-2
de Boor and Cox B-spline papers appear
de Boor’s paper was a Technical Report from the UA Army-sponsored Mathematics Research Center in Madison, Wisconsin and Cox’s paper was a Technical Report from the National Physical Laboratory near London. Both were later published as journal papers, but early distribution to interested parties was crucial
the de Boor Cox algorithm was the first numerically stable and efficient algorithm for computing B-splines
Syracuse 1971-2
Syracuse 1971-2
Rich dreaming of B-splines
Syracuse 1971-2
…in his palatial residence…
Syracuse 1971-2
The power behind the throne
I first met Elaine at Syracuse although I was aware whilst Rich was at Cambridge the previous year that his progress was being followed from afar and not only by Steve Coons
Rich and Elaine argued about the limit case of the closed B-spline when the degree tends to infinity. They agreed that in the limit the curve reduced to a point, but not necessarily a circular point. Steve Coons asked “what else could it be”, and as a fellow mechanical engineer I sided with Steve
Syracuse 1972
Thesis complete but where to go next?
Conspiracy
Where Rich should go is debated
– fortunately Rich makes the right decision
Steve Coons, Bert Herzog and I had correspondence about this and tried to influence the outcome, most probably unbeknownst to Rich (apologies are due!)
Utah!
Dave Evans and Ivan Sutherland
Utah!
The rest is history…
a few months after arriving in Salt Lake City, Rich wrote that he and Elaine had decided that the limit closed B-spline curve was an elliptical point - sometimes I don’t understand mathematicians!