Department of Computer Science and Technology

Technical reports

Motion development for computer animation

Andrew Mark Pullen

November 1988, 163 pages

This technical report is based on a dissertation submitted August 1987 by the author for the degree of Doctor of Philosophy to the University of Cambridge, Churchill College.

DOI: 10.48456/tr-152


This thesis examines the problems posed by the use of computers in the production of animated sequences, and suggests possible solutions to some of them. Over the years increasing emphasis has been placed on the quest for visual realism, with the result that other considerations—such as ease of motion specification—have been overshadowed. Many current techniques put animation in the realm of the scientist programmer rather than the animation artist. This thesis in the main ignores image rendering issues but considers instead the motion specification phase of generating an animated sequence.

The thesis begins by examining the traditional hand animation process and asks whether speed or quality benefits can be achieved by automating parts of it. My own experiences in this area are described based on the design and implementation of KAS, a computer-assisted animation system, which was then used to produce parts of a short animated film for Channel 4 television.

In the light of this experience, other computer animation techniques are considered—beginning with a survey of animation systems developed by other researchers over the years and a discussion of their relative merits. This survey identifies the two techniques in most common use today as being (i) an extension of the keyframing technique used for traditional hand animation, and (ii) a scripting approach, which essentially involves producing a textual description of the desired motion in a specially developed animation language. Both of these methods suffer from serious drawbacks—with keyframing it is difficult to control complex motion, whereas scripting forces artists into a style of working which does not exploit their traditional intuition and experience and is more suited to computer programmers than animators.

In an attempt to overcome these shortcomings, proposals are made for a new style of motion development environment making heavy use of interactive graphics and using computer simulation to guide the motion of the objects being animated. Once suitable simulation rules have been set up, the software becomes capable of dealing with the majority of situations that an object will encounter. This means that the animator need only intervene occasionally in order to steer the animation to the desired conclusion. Two major experiments aimed at determining the usefulness of this idea were conducted: one uses fixed rules in a simple environment (the game of snooker); the other considers a more general realm (cellular automata) and gives the animator the freedom to change the simulation rules at any time.

The conclusion drawn from these experiments is that the proposed method is capable of development to the stage where a powerful tool can be provided for animators to use in a novel but intuitive way—one which exploits their capability as artists and makes only minor demands on them to understand the underlying technology.

Full text

Only available on paper (could be scanned on request).

BibTeX record

  author =	 {Pullen, Andrew Mark},
  title = 	 {{Motion development for computer animation}},
  year = 	 1988,
  month = 	 nov,
  institution =  {University of Cambridge, Computer Laboratory},
  address =	 {15 JJ Thomson Avenue, Cambridge CB3 0FD, United Kingdom,
          	  phone +44 1223 763500},
  doi = 	 {10.48456/tr-152},
  number = 	 {UCAM-CL-TR-152}