Work reported herein was supported in part by Project MAC, an MIT research project sponsored by the Advanced Research Projects Agency, Department of Defense, under Office of Naval Research Contract NONR-4102(01), and by MIT Lincoln Laboratory with support from the US Advanced Research Projects Agency.
This paper is based on a thesis submitted in partial fulfillment for the degree of Doctor of Philosophy at the Massachusetts Institute of Technology, Department of Electrical Engineering.
Animation is the graphic art which occurs in time. Whereas a static image (such as a Picasso or a complex graph) may convey complex information through a single picture, animation conveys equivalently complex information through a sequence of images seen in time. It is characteristic of this medium, as opposed to static imagery, that the actual graphical information at any given instant is relatively slight. The source of information for the viewer of animation is implicit in picture change: change in relative position, shape, and dynamics. Therefore, a computer is ideally suited to making animation possible through the fluid refinement of these changes.[27]
The animation industry is ripe for a revolution. Historical accidents of available technology and knowledge of visual physiology have led to the evolution of the animated film as one that is created frame-by-frame[1]. The prodigious quantities of labor required for the construction of twenty-four individual frames per second of film have led to a concentration of animation activity in the assembly-line environments of a few large companies, an artificial yet rarely surmountable separation of the artist from the medium, and extravagant costs[2]. In conjunction with other trends in American society, the result is usually what the English critic Stephenson describes as the respectable sadism and stereotype of commerce [1]. Yet be offers this hopeful prediction in concluding his 1967 study, Animation in the Cinema: There seems every reason to look forward to changes which would make it possible for the creative artist to put on the screen a stream of images with the same facility as he can now produce a single still picture[1]. This paper explains how a creative artist, aided by a computer, can define a stream of images with the same facility as he can now produce a very few still pictures.
Although the computer's entrance into animation has been a recent one (1964) [3][4], the growth of interest and activity has been phenomenal [5][6][7][8]. Experience to date strongly suggests that the following statements are true:
This paper discusses ways in which man, aided by a computer in an interactive graphical environment, can synthesize animated visual displays. It is widely recognized that such an environment facilitates man-machine communication about still pictures [20][21][22]. The paper seeks to:
Three aspects of the role of direct graphical interaction in computer graphics are particularly relevant to computer animation:
The power of immediate visual feedback in animation is striking. The computer calculates, from its representation of a dynamic sequence, the individual frames of the corresponding movie. Like a. video tape recorder, it plays it back for direct evaluation. A small change may be made, the sequence recalculated, and the result viewed again. The cycle of designation of commands and sketching by the animator, followed by calculation and playback by the computer, is repeated until a suitable result is achieved. The time to go once around the feedback loop is reduced to a few seconds or minutes. In most traditional and computer animation environments, the time is a few hours or days. The difference is significant, for now the animator can see and not merely imagine the result of varying the movement and the rhythm of a dynamic display. Thus he will be led to perfect that aspect of animation that is its core: control of the changing spatial and temporal relationships of graphic information.
Factoring the construction of an animation sequence facilitates the effective use of feedback from ear1y stages to guide work in later stages. Working on individual small subsequences helps overcome the serious practical problems of computer time and space that could disallow rapid enough calculation and playback.
We know from the computer graphics of still pictures that the computer simulates not only a passive recording agent in its ability to retain images, but an active medium which transforms the very nature oft he sketching process. This remark applies trivially to computer animation; one may construct a sequence of drawings to comprise the individual frames of the film, the static images existing at single instants of time. Picture change that extends over entire intervals of time is then synthesized as a succession of individual (temporally) local changes that alter one frame into another.
This paper goes further, for it explains how the computer can be a medium which transforms the very nature of the process of defining picture change, of defining movement and rhythm. Dynamic behavior is abstracted by descriptions of extended picture change. These descriptions may themselves be represented, synthesized, and manipulated through pictures, both static and dynamic. Thus dynamic control can be exercised globally over the entire sequence. What results is one new conception of what it means to draw an animated film.
Figure 1 portrays a suitable environment for interactive computer-mediated animation. Figure 2 is a block diagram. of such a system.
To illustrate the process of animation in an interactive computer graphics environment, we present a scenario. The example, chosen for its simplicity, is an extended version of one actually executed with the GENEralized-cel animation SYStem. GENESYS is a picture-driven animation system implemented on the MIT Lincoln Laboratory TX-2 computer. All capabilities purported to it are operational or could be made so by minor additions. The written form of the interactive dialogue has been adjusted to increase its clarity.
We want to see a dynamic sequence of a dog dashing to his dinner and then dining: The dog runs towards a bowl. Wagging his tail, he lowers his head and laps up the milk. Several slurps of the milk are to be shown before we cut to the next scene.
How we do it:
GENESYS either types or displays this response.
We may distinguish three old approaches to the synthesis of a sequence of frames:
Animation sequences have traditionally been synthesized through the individual construction of frames. The illusion of a continuum of time is attained through rapid playback of discrete instants of time. This approach is the only one applicable to the construction of pictures that defy regular or formal description, and that require unique operations on each frame. Yet the cost is excessive and continues to rise dramatically, faster than the GNP[27]. Salaries in large studio operations typically consume half of the cost, for commercial animation is a complex interaction among producers, directors, designers, layout artists, background artists, key animators, assistant animators, inkers and colourists, checkers, cameramen, editors, and studio managers[2]. It is this division of labor, this dispersal of the creative process, which separates the artist from the medium[27]. Another major weakness of conventional frame-by-frame animation is that there are no efficient methods of making changes to a movie stored on photographic film or video tape. We discuss elsewhere what role the computer might assume in frame-by-frame animation [28].
The technique of interpolation has long been used to cut Costs and reduce the burden of picture construction which is placed on the key animator. Interpolation occurs when the key animator asks his assistants to fill in the pictures intermediate to a pair of critical frames. It has been suggested that part of this process could be mechanized[29]. We do not consider further that problem in this paper.
The generation of a sequence of frames from a formal algorithmic description is a process characterized by:
The form of the expression has to this date been a written program in a picture-processing language such as BEFLIX [3][4] or a sequence of directives in a typewriter-controlled command language such as CAFE[30]. Herein lies another strength of the approach and also a fundamental weakness. On the one hand, many programmers, scientists, and engineers, previously not animators but fluent in this new language, can now produce dynamic displays[31]. On the other hand, an animator trained in traditional media and techniques is forced to learn a completely new language, a completely new way of thinking.
Picture-driven animation is a new process that augments harmoniously the animator's traditional techniques, that reflects and extends the ways of thinking to which he is accustomed. Within his intuitive language of pictures and sketching and mimicking, he may synthesize both components of frames, called cels, and generative descriptions of extended picture change, called global descriptions of dynamics.
Global dynamic descriptions are data sequences, whose successive elements determine critical parameters in successive frames of the movie. Algorithms embedded in a picture-driven animation system combine cels and dynamic descriptions to produce visible picture change. The animator defines and refines pictorial representations of dynamic descriptions. These data sequences then drive the algorithms to generate an animated display. Hence the process is called picture-driven animation.
The process is powerful because it is easy to achieve rich variations in dynamic behavior by altering the data sequences while holding constant a few simple controlling algorithms. The data sequences precisely determine the evolution of recurring picture change, within the constraints set by a choice of controlling algorithms.
We next introduce the three kinds of global dynamic descriptions, some useful algorithms for which they may be driving functions, and some useful methods for their static and dynamic pictorial representation and construction. The following classification will be helpful:
A global dynamic description is either a movement description, which is either a continuous movement description = a path description, or a discrete movement description = a selection description; or a rhythm description.
Consider those alterations of static pictures that consist of modifications of continuously variable parameters, such as location, size, and intensity. Their instantaneous values determine the picture's appearance at a given moment. Thus the static picture may be animated by specifying the temporal behavior of such parameters. A representation of the temporal behavior of a continuously variable parameter is called a path decription.
The movement of a fixed-geometry picture (cel) in GENESYS is described as the change of two coordinates with time, and is represented by a pair of path descriptions. Their specification may be used to synthesize the drifting of a cloud, the zooming of a flying saucer, the bouncing of a ball, or the positioning of a pointer.
Since the behavioral descriptions of the parameters apply to entire intervals of time, the animation is liberated from a strictly frame-by-frame synthesis. The computer is a medium through which one can bypass the static or temporally local and work directly on the dynamic or temporally global. Movement is represented as it is perceived, as (potentially) continuous flow, rather than as a series of intermediate states.
Path descriptions, in fact, all dynamic descriptions, may be defined by one of six general approaches:
We shall in this paper be concerned with techniques implementing the first two approaches only. Sketching is useful when one knows the general shape and quality of a motion rather than an analytical expression for a function that determines it. Modifications of the sketches are frequently invoked after one views the current animation sequence and determines how it is inadequate.
There are two related kinds of pictorial representations of all movement descriptions, static and dynamic. Both kinds may be introduced with a single example.
Consider the motion of a figure that goes from one corner of a square room to the diagonally opposite corner by walking along two adjacent walls. We shall ignore the vertical movement and consider only motion of the center of the body in the two dimensions of the plane of the ground. He first walks in the direction of increasing X coordinate, then in the direction of increasing Y coordinate. We further assume that he begins from a standstill, accelerates and then decelerates to the first corner, pauses there for a brief interval while he turns in place, and finally accelerates and decelerates to his destination.
One complete description of this planar movement consists of the functions of the X and Y coordinates versus time. These are depicted in Figures 7 and 8.
Such representations of changing picture parameters are called waveforms. Time is depicted, in the waveform, along one spatial dimension. The waveform's construction requires movement of the stylus along that dimension; the display records and makes tangible this movement.
Alternatively, both spatial coordinates could denote the two spatial coordinates of the movement. A natural correspondence is established between the X(Y) coordinate of the floor and X(Y) coordinate of the medium of the representation (paper, scope face, etc.). Figure 9 depicts such a parametric curve representation of the movement. It illustrates with clarity the figure's path on the floor.
Yet the dynamics of the motion are hidden because the temporal dimension is only an implicit coordinate. This is rectified in Figure 10. A stream of symbols is used instead of a continuous trail to depict the path. Characters are spaced along the path at short, uniform intervals of time, such as every 24th of a second. Dynamics are apparent in the local density of symbols. Observe in particular how they cluster where the figure pauses.
The dynamic construction of a path description is a user-driven animated display in which the timing of the stylus's movement is preserved by recording its position in every frame. A tangible representation of the stylus path is the display of a sequence of characters spaced equally in time. We shall call a parametric curve dynamically sketched in real time a p-curve. The p-curve corresponding to Figures 7-10 is depicted in Figure 11. We have attempted to convey in a single static image that the p-curve is a dynamic display. Each 2-dimensional p-curve determines two path descriptions. Thus the hopping of the dog in DINNERTIME may be synthesized by hopping with the stylus along some path on the tablet surface, that is by mimicking the desired dynamic.
In some cases one may need only one of the path descriptions. To depict the fluttering of a heart, we may assign the X coordinate of the p-curve to a parameter determining the size of the heart, and then flutter the pen back and forth horizontally. Any vertical motion that results is uninteresting and can be ignored.
A path description, in summary, defines dynamic activity that consists of potentially continuous and arbitrarily fine alterations of value. The reader should not be misled by the choice of the word path. What is meant is a path, or sequence of values, through an arbitrary continuous space, through a mathematical continuum. One application or interpretation of this path is the representation of a movement through the location-space of an object, such as a figure's path through a room. This interpretation, however, is not the only possible one. Depending upon the picture description capability of the system in which it is used, and the algorithm which it drives, a path description may determine changing locations, inteneities, thicknesses, densities, or texture gradients. For example, a pulsating heart could be animated by varying either the size or the intensity of a single heart shape.
Reference 28 presents a detailed discussion of the relative strengths and weaknesses of waveforms, p-curves, and other static and dynamic representations of continuous movement. The discussion focuses on their uses as inputs of dynamics and as visual feedback to the animator, their dimensionality, their role in guiding temporal and spatial adjustments to existing motions, their capacity for conceptual extensions, and some practical problems (and solutions) that arise in the sketching process. Furthemore, we describe four kinds of editing and refining capabilities, operations for:
Consider the algorithm that selects an element of the current frame from among members of a cel class. A good example arises in the synthesis of different facial expressions through the abstraction of discrete shapes and positions of mouth, nose, eyeballs, and eyebrows. One cel cless could consist of the two members eyebrows raised and eyebrows lowered. An animation sequence may be achieved by a temporal concatenation of selections from a cel class. A changing facial expression may be achieved by the parallel application of several such sequences of selections, one corresponding to each facial component. In DINNERTIME this technique was used to synthesize the movement of the dog's legs, tail, head, and tongue.
A representation of the dynamic selection from a finite set of alternative pictures is an example of the second type of global dynamic description and is called a selection description. The synthesis of selection descriptions is also aided by the use of pictorial representations, such as one consisting of a sequence of steps, where the length of each step is an integer multiple of frames, and the height is limited to transitions to and from positions on a discrete scale. Such pictures appear at the top of Figures 15 and 20. Superposition on a common time axis of pictures of several descriptions facilitates coordinating the counterpoint of the parallel selection strands.
The use of the term selection implies that a mechanism chooses from among a designated set of alternatives. In the previous examples the alternatives are cels, images to be introduced as components of frames in a dynamic sequence. A more general view of a selection description regards it as a sequence of selectors, functions which choose from a designated and finite yet potentially denumerable set of alternatives. Depending upon the picture description capability of the system in which it is used, and the algorithm which it drives, a selection description may choose among alternatives that are subpictures, data, picture-generating algorithms, other global dynamic descriptions, pictorial events or activities, or strands of dynamic activity. For example, the dynamic selection from among alternative picture-generating algorithms would be useful in a system with discrete texture choices, where there is one algorithm capable of filling an arbitrary region with that texture.
Further details may be found in reference 28, which also discusses techniques for the definition and editing of selection descriptions. These are conceptually similar to those used in the synthesis of path descriptions.
Rhythm descriptions consist of sequences of instants of display time (frames), or intervals between frames. They define patterns of triggering or pacing recurring events or extended picture change. In this context it is suggestive to think of a rhythm description as a pulse train. Each pulse may trigger the same action, or, as is discussed in reference 28, it may trigger one of several activities under the control of a selection description.
Rhythm descriptions facilitate the achievement of coordination and synchrony among parallel strands of dynamic activity. In this context it is suggestive to think of a rhythm description as a sequence of event markers. The marking sequence may be defined with respect to one pictorial subsequence, and then used to guide the construction of another subsequence.
A rhythm description cannot by itself define picture change; it can define a beat, a sequence of cues with respect to which picture change is temporally organized and reorganized. Animators have sometimes used metronomes as generators of rhythm descriptions[2]. Proper synchronization of a sound track to the visual part of a film is most critical. to its success[2].
Hence, rhythm descriptions marking critical instants of time play a key role in the synthesis and editing of movement descriptions. For these operations a rhythm description requires pictorial representation. In Figure 20 it is depicted both as a static pulse train and as a sequence of event markers along the axis of movie time. A direct and simple dynamic input, as we have seen in DINNERTIME, consists of tapping out the rhythm on a push-button.
It is easy to conceive of more complex and useful couplings of global dynamic descriptions. Suppose, for example, that a hop, a skip, and a jump have each been synthesized with the aid of several path and selection descriptions. If the animator wishes to experiment with varying dynamic patterns of hop, skip, and jump, he should be able to define a selection description which chooses among these three alternatives. This is equivalent to defining selections among sets of path and selection descriptions. Reference 28 discusses the use of selection descriptions to establish arbitrary hierarchies of structured dynamic behavior, and illustrates the significance of this capability to the animator.
Three special-purpose picture-driven animation systems have been implemented on the MIT Lincoln Laboratory TX-2 computer. A common feature is that each has a construction or editing mode, a playback or viewing mode, and a filming mode. In the first mode the animator may begin work on new pictures and global dynamic descriptions, or may recall and continue the construction of pictures and descriptions saved from other sessions. Algorithms embedded in the systems then compute TX-2 display files, in which sequences of frames composed of points, lines, and conic sections are encoded for use by the scopes.
These image files are passed to the playback program, which simulates a variable-speed, bi-directional, video tape recorder. The program normally sequences through the display file representation of successive frames, making each in tum visible for 1/24th of a second. One useful option is that of automatic cycling or the simulation of a tape loop.
When the animator has prepared a satisfactory sequence, he need no longer view it directly on the scope, but may instead want to record it on film. A pin-registered movie camera can be mounted in a light-tight box to a TX-2 scope. Its shutter is always open. The filming program (a variant of the playback program) paints an image on the scope. After a sufficient time interval to allow the decay of the phosphor, approximately 1/5 of a second, a signal from the computer advances the camera. A return signal upon the completion of the advance triggers the display of the next frame. The camera can be operated on one scope while we work at a tablet with another scope. Excellent film quality, with high contrast and low jitter, can be produced with the system.
The first two systems are very special-purpose. ADAM allows one to animate a crude line-drawing representation of a single human figure. EVE is an exercise in abstract dynamic art, in which one can animate a set of points linked by rubber-band straight lines. The animation technique in both cases is the specification, via waveforms and p-curves, of the seventeen path descriptions that define the temporal behavior of the picture's seventeen controlling continuous parameters. A lengthy discussion may be found in reference 28; we shall here content ourselves with three observations:
Although GENESYS is also a special-purpose animation system, it is versatile enough to be used in the generation of a broad class of dynamic images. The term generalized-cel, defined in reference 28, is a generalization of the concept of cel class illustrated in that its appearance in a given frame of the final dynamic display is determined by the values of a set of associated movement descriptions, both continuous and discrete.
The GENESYS animator may sketch, erase, copy, translate, rotate, and scale individual cels consisting of points, straight lines, and conic sections. He may sketch p-curves and dynamically tap rhythm descriptions. There are numerous tools for the manipulation of static representations of dynamic descriptions. Several individuals with varying degrees of artistic skill and training in animation have constructed short cartoon sequences with the aid of GENESYS. Figures 12-20 illustrate some of these experiences.
Thus the essence of picture-driven animation is:
The three kinds of descriptions constitute a rich, expressive, intuitively meaningful vocabulary for dynamics. Each type abstracts an important category of dynamic behavior-flow and continuous change (path descriptions), switching and repetitive choice (selection descriptions), and rhythm and synchrony (rhythm descriptions), The vocabulary is economical, flexible, and general in the sense that it can characterize the dynamic similarities that exist in seemingly diverse animation sequences.
The use of dynamic descriptions couples picture definition by sketching and by algorithm; it furthermore allows both local (of the individual frame) and global (for an interval of time) control over dynamics. We have chosen to stress the latter and adopted the term global dynamic description, for it is the capacity for global control that results uniquely from the use of the computer as an animation medium. Yet a dynamic description is not only a representation over an interval, but a sequence of single elements whose modification also provides local control over individual frames Both local and global control are vital to the successful synthesis of movement. He who accidentally crashes into a wall while running from the police is going from the continuous to the discrete, from a global motion to a local event. He who aims to scale the wall is interpolating the continuous between the discrete, adjusting the global to fit the constraints of the local.
The naturalness and power of the vocabulary is increased by the ability to manipulate it in an interactive graphics environment. There exist, for each kind of data sequence, static pictorial representations such as the waveform which provide a global view of and facilitate precision control of the temporal behavior implied by the sequences. There exist, for each kind of data sequence, methods of dynamic specification such as the clocked sketching of parametric curves which allow the animator's sense of time to be transmitted directly through the medium of the computer into the animated display.
We use the term global dynamic description and the names of the three types somewhat loosely in referring both to the underlying dynamic data sequences and to their corresponding pictorial representations. The imprecision is purposeful, for it is very significant that, in an interactive graphics environment, one can easily traverse in either direction any leg ot the triangle {Dynamic Data Sequence, Static Pictorial Representation, Dynamic Pictorial Representation}. What results is an important plasticity in the representation of dynamics. Characterizations of change can be manipulated (shifted, stretched, superimposed, ... ) within and between the domains of the static and the dynamic.
Several animation sequences can readily be related, coordinated, or unified, regardless of whether or not they ever occur concurrently. Dynamic behavior (data) can readily be transferred from one animation subsequence (including the animator) to another, from one mode of representation or embodiment in a picture to another.
Our concept of a picture is a broad one, and purposely so. For as we stress in reference 28, a computer-mediated picture is not only what is visible but what is contained in its model in the computer system. And the system, i.e., an interactive animation system, includes not only disks and core but an animator and perhaps an ongoing physics experiment as well as a tape-recorded speech. This system evolves continually through real time. Occasionally there occurs a particular reorganization of the system which results in the transfer of information from the animator to the pictorial data base, or in a computation on the data base which results in a sequence of visual images (i.e., data directly convertible by hardware into visual images). Thus, as we have stressed before, the act of mimicking dynamics is a (user-driven) dynamic picture. This unification of the concepts of picture and action is important.
The greater is the number and generality of available models of pictures and of processes of picture construction, the more flexible and powerful is the animation system in its ability to deal with dynamic information. The design of a multi-purpose, open-ended animation language that, allows the animator himself to synthesize new models is outlined in reference 28. With such a language one can describe arbitrary action-picture interpreters that extract movement descriptions from the animator's use of system devices and transform them and existing static and dynamic displays into new static and dynamic displays.
Finally, the use of dynamic descriptions helps establish a conceptual framework which facilitates efficient use of the resources of the animation system: animator, software, and hardware. For details, we again refer the reader to reference 28.
This paper is a pointer to a March, 1969, Ph.D. dissertation[reference 28] which includes the material contained herein considerably expanded, some suggestions for future research, and
With respect to the arts, we conclude by repeating McLaren's description of animation:
Animation is not the art of DRAWINGS-that-move but the art of MOVEMENTS-that-are-drawn.
What happens between each frame is more important than what exists on each frame
.Animation is therefore the art of manipulating the invisible interstices that lie between frames. The interstices are the bones, flesh and blood of the movie, what is on each frame, merely the clothing.
This paper may be regarded as a report on a use of the computer in the art of MOVEMENTS-that are-drawn, in the manipulation of the invisible interstices that lie between frames.
The encouragement, counsel, and insight of the dissertation's mentor, Professor Edward L. Glaser of Case Western Reserve University, and of Dr. William R. Sutherland of MIT Lincoln Laboratory, Professor Murray Eden of MIT, and Mr Eric Martin of Harvard University and Cambridge Design Group, Inc. are gratefully acknowledged. We appreciate the support of numerous individuals, here nameless but not forgotten, many in the Digital Computers Group of MIT Lincoln Laboratory, who have contributed to the progress of this research.
1 R STEPHENSON Animation in the cinema A Zwemmer Limited London A S Barnes and Co, New York 1967
2 J HALAS R MANVELL The technique of film animation Hastings House New York 1959
3 K C KNOWLTON A computer technique for producing animated movies Proc SJCC 1964
4 K C KNOWLTON A computer technique for the production of animated movies Bell Telephone Laboratories Film
5 The human use of computing machines Bell Telephone Laboratories Symposium June 20-21 1966
6 Conference on Computer Animation Education Development Center Newton Mass July 17-18 1967
7 Proceedings of the 1967 UAIDE Annual Meeting
8 Proceedings of the Fall Joint Computer Conference 1968
9 F W SINDEN Force, mass, and motion Bell Telephone Laboratories Film
10 J L SCHWARTZ E F TAYLOR Computer displays in the teaching of physics Proc FJCC 1968
11 MIT SCIENCE TEACHING CENTER Scattering in one dimension Film available on loan from the Atomic Energy Commission
12 C LEVINTHAL Molecular model-building by computer Scientific American Vol 214 No 6.June 1966
13 C LEVINTHAL Computer construcion and display of molecular models Film
14 E E ZAJAC Computer-made perspective movies as a scientific and communication tool Comm ACM Vol 7 No 3 March 1964
15 E E ZAJAC Two-gyro, gravity gradient attitude control system Bell Telephone Lahoratories Film
16 S VANDERBEECK J H WHITNEY Several animated films made with the aid of a computer
17 Design and the computer Design Quarterly 66/67 Walker Art Center Minneapolis Minn
18 A M NOLL The digital computer as a creative medium IEEE Spectrum October 1967
19 J REICHARDT Cybernetic serendipity, the computer and the arts Studio International London and New York 1968
20 J C R LICKLIDER Man-computer symbiosis Trans IRE PGHFE HFE-1 4 1960
21 I E SUTHEHLAND Sketchpad: a man-machine graphical communication system MIT Lincoln Laboratory Technical Report No 296 Jan 1963 Proc SJCC 1963
22 J C R LICKLIDER Man-computer partnership International Science and Technology May 1965
23 M A DAVIS T O ELLIS The rand tablet: a man-machine communication device Proc FJCC 1964
24 J F TEIXERA R P SALLEN The sylvania data tablet Proc SJCC 1968
25 L G ROBERTS The lincoln wand Proc FJCC 1966
26 J E CURRY A tablet input facility for an interactive graphics system Proc of the International Joint Conference on Artificial Intelligence 1969
27 E MARTIN Private Communication
28 R M BAECKER Interactive computer-mediated animation Ph D Dissertation Department of Electrical Engineering MIT March 1969
29 T MIURA J IWATA J TSUDA An application of hybrid curve generation - cartoon animation by electronic computers Proc SJCC 1967
30 J NOLAN L YARBROUGH An on-line computer drawing and animation system IFIPS 1968
31 W H HUGGINS D R ENTWISLE Exploratory studies of films for engineering education Department of Electrical Engineering The Johns Hopkins University Report to US Office of Education September 1968
32 W R SUTHERLAND J W FORGIE M V MORELLO Graphics in time-sharing: a summary of the TX -2 experience Proc SJCC 1969
33 W H HUGGINS D R ENTWISLE Computer animation for the academic community Proc SJCC 1969
34 N MCLAREN Quotation in animation exhibit in the Canadian Cinematique Pavilion EXPO '68 Montreal National Film Board Canada