Computer Animation: Answer or Problem?
Alan Kitching, Grove Park Studio
December, 1971
British Kinematography Sound and Television
There have been several recent opportunities to see computer-made films at Society's own lecture programme. Such increasing attention reflects the rapidly advancing sophistication of computer animation systems, and this article examines some of the more important of these recent developments.
THIS picture is from a one-minute cartoon called The Apteryx and the Easter Bunny. The animation is of a standard at least as good as the average quality of animated films we see on TV. What is remarkable about it, though, is this - every scene in the film took no longer to make than it takes to project; in other words, if Easter Bunny walks across the screen in 4 seconds, it takes just 4 seconds to make that piece of film ....
The device that produced the film is a machine called Caesar, which is one of a range of computers designed and built by the Computer Image Corporation of Denver, U.S.A. We will look at their work in more detail later in this article; the reason I want to start with this example is to emphasise one point - many people dismiss computer animation as good only for abstract patterns and scientific diagrams; yet the Apteryx film, which was made over a year ago, dramatically demonstrates how far these limitations have been overcome, and for this reason the film has already become something of a historical landmark.
Attempts to make forecasts of technical progress are frequently branded as over-optimistic, but it is surprising how often the record shows the error as entirely the other way. In 1969 it was quite daring to predict that computers would be capable of instantaneous full-colour character animation within a decade; but two years later it has already been achieved. The prediction was wrong 400%. The reason for this spectacular error is quite specific and technically significant: up till two years ago, almost everything produced under the label of computer animation was produced on computers that were (a) digital, and (b) designed for some quite different purpose - and for systems of this kind, the prediction could still be not far off the mark; what has made the difference, though, is the emergence of an alternative technical approach resulting in computers that are (a) analog, and (b) designed from the outset with animation as the sole objective. The technical distinction between a digital computer and an analog computer is therefore quite crucial to a full appreciation of the progress that is being made; for the moment, however, I want to leave this topic, and concentrate on describing some of the most significant advances in the traditional digital approach.
The idea that anything to do with computers could be described as traditional may seem surprising, but it accurately indicates how the technical horizons have widened. However, I do not mean to give the impression that the digital computer is now obsolete, or in any fundamental way inferior. On the contrary, the digital approach may well win out in the long run; at present, however, computer power is so expensive that there are many limitations on what can be achieved; for the same reason, of course, almost all the pioneer work is being done in the United States.
In this country, one of the few major centres of activity in this field is the Science Research Council's Atlas Computer Laboratory at Chilton, Didcot, where a one-day Computer Animation Symposium was presented on July 30. The major part of the symposium programme was devoted to discussing various aspects of work produced at the Atlas Lab itself; and since it is not unfair to describe the Chilton installation as traditional computer animation, it makes a convenient starting point for this survey.
The facilities at Chilton presently consist of the Atlas computer, a Stromberg-Carlson Datagraphix S-C 4020 microfilm recorder, a D-MAC pencil follower, and an Algol software package called GROATS; for the future, more powerful computers and more sophisticated software are being planned and installed.
The device that actually produces the image is the microfilm recorder; at the heart of this is the Charactron tube, which is a specialised form of cathode-ray tube. The electron beam of this CRT can be shaped into any one of a number of standard alphabetic characters, and directed to a desired location on the screen. Lines can be drawn by selecting a dot and causing it to move from one specified point to another. The screen is divided into a grid of 1,024 rows and columns, and a point is specified by a pair of numbers which are its co-ordinates on this grid. Images displayed on the screen are photographed directly by a more-or-less conventional film camera.
Making a film with this apparatus is a fairly lengthy procedure. The image on every frame has to be specified precisely in terms of all the little bits of straight line which are needed to draw it. There could be millions of these in a complete film, so the task would be unthinkable for an unaided human. The job of reducing this task to manageable proportions is accomplished by the Atlas computer and its GROATS software; the trick that makes it work is that any simple and repetitive mathematical routine can be programmed into the machine and given a special code-name. To draw a circle. for example, it is not necessary to specify all the points on the circumference; by using the appropriate codeword you need only specify the radius of the circle and the location of its centre, and then the computer will calculate all the required co-ordinates. This means that now we need only specify each frame of film in terms of the geometrical shapes which are needed to draw it. The trick can also be carried further; we could draw a face composed of two circles for eyes, a triangle nose, a rectangle mouth, all contained in a large circle, and give it the code-name face. However, the result would look obviously geometrical; and computer power is so limited that only figures that are mathematically fairly simple can be easily animated.
Once the specifications for the images have been prepared, they are fed into the computer which then calculates the co-ordinates of every line segment needed to draw the film. This information emerges as a magnetic tape; the tape is then loaded on to the microfilm plotter, which generates the film. If you want colour, it must be added afterwards by optical printing through filters.
There are two obvious reasons why computer systems of this kind have failed to make any spectacular impact on the everyday world of animation and graphic design: they cannot produce the kind of images that designers want; and they need specialised technical knowledge to operate them, which few designers are prepared to learn. Also there are problems of cost and accessibility, but these could probably be solved if the first two problems didn't exist.
However, there is one further thing that Atlas and the other traditional systems can achieve. So far, I have said that what the computer does is to translate the programmer's geometrical specifications into detailed electronic instructions for plotting lines. But we can also use the computer's mathematical ability to decide some of the specifications in the first place. To take a simple example, suppose that for some reason we want to animate the flight of a cannon-ball. We could choose a scale, such as one grid-unit equals one foot; at this scale the width of the screen would represent a distance of 1,024'. In the bottom left-hand corner we might draw a little cannon, perhaps with a rectangle for a barrel, and circles for wheels. First we would have a number of static frames as an establishing hold, then at a specified frame the cannon-ball would appear, represented by a small circle. Suppose we have drawn the barrel tilted upwards at 45°; if we now specify that the cannon-ball is to emerge travelling at a speed of (say) 175 grid units per second in a direction 45° to the horizontal, and inform the computer that the ball is to suffer a continuous acceleration vertically downwards of 32 grid units per second per second (representing gravity). the computer has all the information necessary to calculate the trajectory of the cannon-ball. We then ask it to calculate the positions corresponding to each successive interval of 1/24th of a second and to draw the ball in whatever position this calculation determines. The result is an eight-second sequence that accurately simulates the physical reality it represents (apart from the fact we have ignored air resistance).
This ability to simulate physical processes by instructing the computer to apply given laws to determine the disposition of picture elements is a facility unique to computer animation, and is therefore a totally new tool. Simulations of this kind are undoubtedly the most successful area of application for traditional computer animation, and many films of this kind have been produced that simply could never have been made in any other way. However, the topic of simulation is being covered in more detail by Ian Duff in a separate article in this issue, so I want now to turn our attention back to the problem of building computers for animators and graphic designers, and to the developments that have made Atlas seem frustratingly antique.
The next major step forward in digital computer animation is the development of interactive systems. These systems display the results of an instruction instantly on a CRT screen in front of the user. Instructions may be fed in by one or more of three methods: (a) typewriter keyboard; (b) special-purpose function keys, or (c) by using an electronic light-pen which can be used either to draw a line directly, or to identify a particular part of the image which the user wishes to alter in some way. With this development it is no longer necessary to programme the entire film in advance, since the user can work directly with the image. He can build it up piece by piece, modifying it as he goes along, until finally he gets the result he wants. Also, the user needs no specialised technical knowledge, other than a familiarity with the machine's controls, so the first of the two major barriers between the computer and the designer is overcome. Several examples of interactive digital systems were described and illustrated at the VIS-COM '71 congress by Patrick Beatts of IBM, and this presentation convincingly demonstrated how easy it is to use such systems.
One of these examples was a film called Genesys-1 made by Ron Baecker at MIT Lincoln Laboratory; the equipment is a TX-2 computer with CRT and Sylvania light-pen tablet, and Genesys is the name of the program. The film shows in detail the steps followed by an animator to animate a bouncing ball. Images are drawn on the tablet with a light-pen, and simultaneously appear on the screen. Parts of the drawing can be instantly erased by pointing at them with the light-pen and pressing an erase key. Finalised drawings can be stored and recalled at will by pressing keys. The light-pen can also be used to describe motion, by drawing what is called a p-curve, The following diagram illustrates the principle:
The p-curve is what animators call the path of action; when the animator selects this function, the light-pen no longer draws a continuous line but instead fires dots at a regular rate of (say) 24 per second: thus when the animator draws a line, it appears as a line of dots whose spacing exactly corresponds to the speed of movement as well as the direction taken - the closer the dots, the slower the movement. It's rather like a machine-gun firing bullets. Any stored image can then be instantly fitted to these positions. The illustration here shows all the positions superimposed, but the computer can put them on successive individual frames and show them instantly at normal projection speed. This normal-speed facility is enormously useful - it is known by the term real-time, which we shall be using frequently in the rest of this article. The animator can immediately see the effect of his animation; if he wants to speed it up or slow it down, this is easily done; if he wants to alter specific frames - such as the ball's bounce - he can recall it and make the correction. When he is finally satisfied, he can hit the record button and the whole sequence is recorded - in real-time.
Patrick Beatts also showed us a film called Sketchbook made by Charles Csuri at Ohio State University. This uses an IBM 1130 with an IBM 2250 CRT display unit, and takes us another step forward. The method of working is similar to Genesys - light-pen and function keys. The computer has a similar ability to move drawings along selected p-curves - these may be either drawn freehand, or mathematically defined. But more than this, it can also manipulate objects in perspective - for example, if you feed in drawings of an object viewed from various sides, it can fill in all the inbetweens needed to animate the object rotating. This is quite a remarkable advance when you consider the pitfalls that can arise, such as the rotating beer-can problem (which arises when lines representing the boundaries of curved surfaces have to be drawn), and even more difficult, the hidden line problem, both of which are illustrated here.
The hidden line problem is a particularly difficult one for computers to solve. It is a comparatively simple matter to get a computer to draw a perspective like drawing (a) where all edges of the object are visible, as if it were a wire framework or a perspex model. If we wish to represent it as a solid object, however, we must regard the lines as boundaries of opaque surfaces, and surfaces nearest to the viewpoint will hide the boundaries of ones that are further away; also it is a fact that all transparent drawings like (a) above are ambiguous - they can always be seen in two different ways - (b) and (c). The problem of removing hidden lines automatically on the computer is therefore quite complicated, and requires a lot of computer power to solve. The system at Ohio State University can do this for certain kinds of drawings, and can do it in real-time. In addition, drawings can be separated into segments that can be made to move independently - for example, the Sketchbook film contained a sequence of a helicopter, which was made to fly round the screen, in all directions and angles, with its two rotors spinning - all in perfect perspective, and all controlled directly by the animator in real-time. The helicopter moved against a night-sky over a city; this was a photographic background that was added afterwards by an optical printer. This system is also capable of painting surfaces with solid tone; at the moment, however, the computer is rather slow, and this can't be done in real-time.
All the computer systems so far described have been only moderately expensive, and the results produced are all economically reasonable. The ideal of computer animation is that eventually we want to be able to do anything we like with any image we like. The image output for an ideal system would therefore have a random complexity equal to that of a colour TV signal. It is fairly easy to show that to handle this digitally (as in pulse-code-modulated TV) the information rate needed would be of the order of 100 million bits per second, which is several sizes too big for any existing system. However, there seems to be virtually no theoretical limit to the size of computers in the future, and we have every reason to expect that the next 30 years really will produce computers as powerful as Kubrick's notorious Hal; on this scale we can expect ideal digital computer animation within 20 years, and perhaps even within 10. If that sounds over-optimistic, then I would remind you of my opening remarks, and recommend you to see the film I am about to describe.
This is my last and most sophisticated example of digital computer animation. It is a much more expensive system than any I have already described; it was originally used to simulate moon-landings as part of the space-programme, and perhaps this is why sums of money were spent that would not usually be available. However, computer costs are coming down all the time ....
The system uses an IBM 360/91 and a Fortran graphic program called INTU-VAL, and was developed at the University of California, Los Angeles, jointly by the University and General Electric, under the direction of Peter Kamnitzer. Its purpose is the real-time dynamic simulation of environments rather than the making of films, so the films that have been made are really passive records of a live process. One of these films, called Hancock Airport, was shown at the Chilton Symposium. What happens is this: a real or imaginary environment is selected and fed into the computer as a set of plans and elevations, together with colour specifications for all surfaces. The operator sits in front of a colour CRT screen and keys in any desired viewpoint. The perspective view from this position is calculated, hidden lines removed, and the result is coloured and presented on the screen as a picture as realistic as any Disney could draw - and all in 1/30th of a second! From this point on, movement can begin. The operator has a joystick control which allows him to move his viewpoint at any speed in any direction, including the vertical. The response of the joystick can be varied to simulate, for example, a light aircraft, or a jumbo-jet. Hancock Airport is a ten-minute record of such a flying simulation, and includes taxiing, take-off, circling and landing. The climax comes when our imaginary aircraft is circling the airport and suddenly the sun sets, the sky goes dark, and the airport lights come on; the aircraft then makes a perfect night-landing. The image quality is so good that it's sometimes hard to remember you're not looking at a live-action film, yet you know that all this really exists only as a set of numbers. The nightfall effect has enough dramatic quality to provoke oohs and ahs - in much the same way as Lumiere's primitive film of a train affected Victorian audiences.
To sum up, then, we can see that the ideal digital animation system is not so very far away - the end of the seventies prediction is still reasonable, perhaps even conservative - but the story doesn't end here: we must now look at what might be considered a basic flaw in all digital approaches to animation. Almost all computer animation has been conceived upon pre-existing digital machines originally designed for quite different purposes; it has been a case of here is a working computer system - how can I make films with it? As we have seen, if the film is mathematically fairly easy to describe or to define, then this approach can be very successful. But what would happen if our approach was to ask we are animators - how can computers best work for us?? A possible answer is that it can be applied to automate the tedious routine work of the animation studio - for example, computer-controlled rostrum cameras, costing, scheduling, and perhaps even computer-controlled puppets. Several such computers exist and are proving successful. In this article, however, I am excluding this approach and concentrating purely on the computer-generation of animated images. Since I am trying to convince you that the days of traditional cel-animation are definitely numbered, it would follow that computerised cel-animation is only of interest as an interim stage with a limited future; it could be seen, therefore, merely as a temporary bridge to another technical world. So I would modify our question to how can computers best generate images?
Since the work of animators and designers is often far from being mathematically easy to define or to describe, the digital computer would hardly commend itself as obviously ideal. The designer needs to be able to create images that are complex, textural, and mathematically arbitrary. He needs to be able to work directly with the image, and without having to become a programmer. On the other hand, since he deals with images that are generally symbolic rather than realistic, complete mathematical precision is not usually needed. In other words, animation is not concerned with creating a faithful simulation of photographic reality, but rather it concerns the creation of an imaginary version of reality; and the imagination does not require such precise information. In a sense, therefore, the digital computer is a 20-million-dollar solution to a 2-dollar problem - rather like the cost of blowing up a bridge in VietNam. This is why the graphic standard of digital films is usually agreed to be very low - practically the only exceptions being the work of Bruce Cornwell and some of the abstract film-makers such as the Whitneys, and Kenneth Knowlton and Lillian Schwartz. At the Chilton Symposium, Tony Pritchett described his work in making digital computer films for the Open University. In an attempt to improve typographical standards, he devised a program that would replace the ungainly geometric numerals with Helvetica - a typeface recognised to be of the highest-quality design. Unfortunately, however, this simple move proved to be disproportionately extravagant.
Also at Chilton, Judah Schwartz of MIT, a pioneer of digital animation, made a similar point when he explained his greatest problem : with digital animation you get too wrapped up in unnecessary and tedious precision. He pointed out that the human brain is a parallel information system combining many different activities simultaneously. Perceptions are relative: we generally know when we want to turn the dial up a bit and we don't care what the reading is. These are characteristics of analog systems, and so give good reason for suspecting that analog computer principles may be more appropriate to the generation of images for visual communication. In a digital computer you basically have one kind of circuit which reduces everything to simple terms of yes/no and tackles the problem one step at a time. This process has great precision, but the needs of the graphic artist are such that an enormous number of steps are required for even the simplest problem. An analog computer, however, has a large number of circuits of many different kinds, and they can all work on a problem simultaneously. Also, information is not broken down into digital black/white terms, but is handled as a varying waveform of all shades of grey. Thus the analog machine is a parallel processor working with relative values - just as we described human perception a moment ago.
The first computer company to take this approach to animation was formed just two years ago, and is the Computer Image Corporation; we have already seen an example of their machine's output at the beginning of this article. The basic principles, however, were actually worked out by their Chairman, Lee Harrison, over a period of 14 years. Actually, there are three different computers- Scanimate, Animac and Caesar, and a fourth, Animac II, is either just completed or is virtually so. Also, it is not strictly correct to call them analog computers, since they do contain digital elements. The two types of computer are not incompatible, and since there are advantages and disadvantages for each, it makes sense to combine them in whatever way is most appropriate to the problem. Such a combination is called a hybrid computer; in the Computer Image machines, analog principles are used for image-generation, and digital principles for data storage.
Scanimate is the smallest of the computers, and is reported to be capable of manufacture for as little as $25,000 - less than the price of a good VTR machine. It consists firstly of a near-normal B&W TV camera that accepts all kinds of graphic input - such as drawings, typography and photographs - and feeds it into a hybrid central processor. The designer can operate the computer directly via manual controls on the computer console, and the results are instantly displayed on a B&W CRT screen. With these controls, the images from the camera can be divided into separate areas, each of which is described by a separate set of analog circuit elements; each analog may then be independently subjected to a bewildering profusion of electronic displacements and transformations, and all can be combined simultaneously on the screen. Finally, the B&W image is fed into a colour-synthesiser, which is basically the same as the Philips Colour Organ system that was shown at VIS-COM '71. That is to say, it separates the image into areas corresponding to up to five possible levels of grey, and each level may be independently painted in any desired colour. The final coloured result appears on a second TV display. Needless to say, the whole system responds in real-time, and once the designer has achieved a satisfactory result, he merely presses a button and the whole thing is recorded.
So here we have a machine that overcomes both of the technical barriers that have kept computer animation off TV and cinema screens; and since the cost of Scanimate productions can be as little as 10% to 30% of the cost of conventional animation, the economic barrier is shattered as well. And not only that, but the production time is also slashed by a similar percentage! However, although the range of Scanirnate's effects is large, they are mainly of the stretching-twisting-exploding type that appeals principally to graphic designers, so it is mainly in the area of titles, special effects and commercials that Scanimate finds its market. And in this market - not surprisingly - it has been extremely successful. Scanimate studios have opened in New York and Hollywood; like Computer Image does it has become a standard phrase among Art Directors, and Scanimated sequences pop up ubiquitously. In this country, Computer Image diagram sequences have recently been seen on the controversial American kids' programme Sesame Street.
Animac takes things a step further by adding the capacity for generating line drawings electronically to Scanimate's ability for working from photographic input. Animac's drawings are created by the manipulation of dials, but the results are similar to digital light-pen drawings; and although Animac is limited to two-dimensional drawings, these can be moved, turned and twisted in full three-dimensional perspective. Anirnac's first commercial film was called Growing and was produced for Encyclopedia Britannica as an educational demonstration of the principles of growth processes. No external artwork was used - all the images were generated electronically. The film is 8½ minutes long; yet the entire production was created in three days, and final filming took just 8½ minutes. By conventional methods, it would have taken at least three months and would have cost ten times as much. And the standard of the resulting film was sufficient to win it the Gold Hugo award at the Chicago Film Festival.
Further vistas are opened up by the fact that other kinds of input besides manual control can be used to manipulate images. Any kind of energy input can be applied, in fact - for example, sound. Music or speech can be used to animate an image in perfect synchronisation. And if the image we wish to animate is a character drawing, we can link the speech input directly to the character's mouth to produce perfect lip sync in realtime. This technique has been developed to such a degree of accuracy that it can be used to help teach languages, or to aid speech-retarded people. One more form of energy input is the anthropometric harness. This is a lightweight harness that is strapped to the body, but is designed to avoid any constriction of movement - its function, in fact, is to measure the movement of each limb and relay this information to the computer.
The computer relates each limb of the harness to the corresponding limb of the drawn character, so that whatever movement the harness-wearer makes is instantly duplicated by the character on the screen. Of course, the harness could also be fitted to forms other than human, and the image need not be a humanoid character ....
Next step up is Caesar, which combines all the capabilities so far described, and includes the ability to paint-in drawings and superimpose them, with hiddenline and hidden-area removal, on any background. This is the first device to achieve character animation in full colour, perfect lip sync, and real-time motion. Apteryx and Easter Bunny was the system's first test.
Finally, we come to Computer Image's latest and most sophisticated system, Animac II. This machine adds the capability of full three-dimensional perspective animation of complex images, such as humanoid characters. In an article in a recent issue of International Photographer, George Toscas reports Lee Harrison as categorically stating that Animac II will have the quality of a Rembrandt in texture, colour and lighting, plus all the character subtleties of a Disney Snow White. Also included is a development called hierarchical programming, whereby specific types of character motion - running, skipping, or marching, for example, can be separately stored and used to animate any character. Before long, there will be a library of such functions.
Well, we have come a long way from patterns and graphs; and the first conclusion to be drawn is a warning to stick-in-the-mud animators: Computer Image is a commercially-oriented business, and it's knocking on the door. It can probably animate anything you can, and a lot more besides - but it does it maybe ten times cheaper and faster than you can do it by hand. It makes no sense to imagine that cel-animation can survive this kind of competition for very long. In not very many years from now, cel-animation is likely to become a wondrous rarity, like an obscure folk-art. But then, to some people, animation has always been rather like an obscure folk-art ....
However, there are much more important conclusions to be drawn, if only we can disentangle the complex jumble of evidence. We want to be able to answer questions like the following: (a) is computer animation a revolutionary new medium, or merely a new technique for an old medium? (b) will it be a Good Thing = Communication and Enlightenment, or a Bad Thing = Silly Cartoons and Flash Graphics? (c) is computer animation the answer, or the problem?
The most fruitful method of tackling big general questions of this kind is to examine the evidence in terms of the sorts of cultural perceptions involved - by which I mean the unconscious habitual assumptions we use in order to comprehend and interact with our environment. This is the same thing as Thomas Kuhn's concept of a paradigm which I described in my article on VIS-COM '71 in the October issue of this journal. The article attempted to demonstrate that the paradigm concept is an invaluable tool for the resolution of such questions, and for this reason I shall use it again here.
Now to the evidence. On the one hand we have digital animation, hitherto largely concerned with physical simulations and abstract geometrical images. On the other hand we have Computer Image working hard to make it possible to re-create Snow White in three weeks. Does that cover all the possibilities, or have we missed the point somewhere? Animac is very like a visual equivalent of a Moog synthesiser. Now a Moog synthesiser can be used to give an excellent performance of Bach, but it can also do many more quite undreamt of things besides. An exactly parallel criticism can be made of Computer Image's concern to perform Disney. In fact, the same criticism might also be applied, but less strongly perhaps, to digital simulations and abstractions, since these could possibly be seen as new and more complex versions only of existing ideas. More often than we may realise, therefore, computer animators resemble the Generals Who are Always Ready to Fight the Last War. But if Kuhn's concept teaches us anything at all, it is that this situation is the normal and inevitable prelude to a revolution of some sort. If you recall, a revolution is defined as the process which occurs when an existing paradigm runs into paradoxical difficulties and is gradually replaced by a new and more appropriate one. In the previous article we also examined evidence which suggests that a large-scale revolution is indeed under way, but that the new paradigm has not yet emerged, so we are therefore still in the paradigm-crisis stage. If you also recall that paradigms define the nature of our perceptual habits, it should follow that any new tool which affects our image-making capabilities (and therefore our visual environment) is inevitably going to be involved in the emergence of the new paradigm and its revolutionary perceptions. In what ways, then, might the various forms of computer animation contribute to the shifting of perceptions?
Simulation can clearly make a significant contribution; as Judah Schwartz pointed out, its effect is like a microscope and telescope combined, since it is capable of showing us events whose scales are otherwise unimaginably too large or too small to comprehend. The computer gives us a kind of intellectual lever with which we can extend our perceptions into other ranges of space and time. Such extensions would tend to foster a somewhat different and enlarged perception of reality to the traditionally normal one.
Mathematically-based abstract animation too has a contribution to make. John Whitney makes the point that computer animation based on patterns of harmonic relationships between periodic elements is a close visual equivalent of musical composition - this new instrument has initiated into the world of the visible certain periodic phenomena that have previously belonged to the domain of music. It is this unique power of the computer to deal with complex periodic forces in terms of motion that I would hold most significant. The emergence of a time-oriented visual art of this kind would also tend to foster a transformed sense of reality by creating a new way to perceive the underlying structure of our environment.
Lastly, we have to consider Computer Image and the new animator-oriented digital systems, which appear to be aimed at producing traditional kinds of animated images, but more quickly. This is an even more blatant case of Generals Fighting the Last War - but the next war always turns out to be different from the last, and especially so when the last war was a defeat. By this I mean to imply that traditional cel-animation has generally failed to achieve something of which it is theoretically capable. To elucidate this, it is necessary to question the fundamental values and assumptions of the animated image.
Almost without exception, sensitive animators have long been aware of a vague dissatisfaction with our depth of understanding of the medium. It is common to hear remarks like John Halas's: animation is a completely unexplored art. The potential is so vast and so little has yet been achieved ... . Two of the speakers at VIS-COM '71 made brief comments on this problem - Dusan Vukotic (Yugoslavia) and Sidney Goldsmith (NFBC) - both of them animators. Vukotic described four reasons why animation frequently fails to communicate as effectively as it might: (a) failure to understand the relationship between verbal and visual knowledge; (b) failure to eliminate superfluous associations; (c) failure to grasp the principles of redundancy and reinforcement; (d) failure to understand that movement conveys a wealth of hidden associations. Sidney Goldsmith pointed out that animation is a means of discovering the essential nature of the moving image, and (quoting Norman McLaren) that what happens between frames is more important than the frames themselves. In motion, a simple dot can express joy or sorrow, and the laws of the real world are suspended. The unexplored potential of animation lies in the fact that it deals with the subjective creation of imagined actions, as opposed to live-action, which involves the objective re-creation of real actions. Animation offers generalised, symbolic images conveying meaning in a compressed, mythical manner operating simultaneously at several different levels. Why has animation not developed this potential?
The obstacle is our imagination. The paradigms we hold today do not recognise these subtler levels of perception, and the perceptions we do hold are those appropriate to live-action; genuine examples of real graphic insight are rare, and the laborious nature of conventional techniques has largely excluded any chance for random exploration. Therefore, although superficially the styles are varied, the underlying perception is still rooted in the cinematic assumption of photographic realism. This is what I meant by saying the Generals' Last War was a defeat - we still do not understand the potentialities of the animated image; and to achieve this, a new paradigm is necessary.
VIS-COM '71 revealed how serious the communications crisis is; and there was general agreement that it could not be resolved with traditional media, but that co-operation between designers, scientists and educators would produce more sophisticated media capable of resolving the problem. However, there is a paradox here: if we expect solutions to communications problems to come from such co-operation, we must prepare for disappointments and delays, since successful co-operation between specialists will require the very tools we are asking them to invent. The barrier is the inability of the specialist to communicate his own paradigms.
To sum up, therefore, we have a communications crisis - which is, incidentally, an integral part of a wider cultural crisis. This can be resolved only by the emergence of a new paradigm, which will necessarily involve new modes of communication and new perceptions of reality. Theoretically, animation has the potential to handle these things, but so far it has not been able to break through the barriers of the old paradigm, because the technical limitations have effectively prevented random exploration on a large scale. Computer animation promises to make large-scale random exploration possible, particularly when you take into account certain other parallel developments - which space allows me only briefly to mention - such as computer terminals for domestic communities, and computer cassette libraries The paradoxical nature of paradigms is that they cannot be explicitly stated; only in retrospect can they be seen to have arrived. This is why random exploration on a wide-scale is the only method by which new paradigms can be generated. And this is also why the kind of approach to computer animation most likely to solve this problem is the approach taken by Computer Image - in spite of being the one apparently most dedicated to the old tradition. Computer Image is closest to giving us the freedom to do anything we like with any image we like, and cheaply; and since we cannot tell in advance what the new paradigm will be, nothing short of this will do. Of course, digital techniques may eventually catch up and surpass Computer Image - alternatively we may find the two approaches converging, if Computer Image increases its use of digital elements and digital systems go hybrid. Be that as it may, however; can we see any clues which might guide our explorations in the use of these machines? I want to close this article by offering two suggestions, though many others are possible.
The first is the direct approach of Ota Yukio, of Japan, who described his work at VIS-COM '71. He has spent seven years developing an international visual language called LoCoS. It uses a few standard elements to build up picture words; these are disposed according to simple rules, so that with very little study, anyone can grasp the meaning. It therefore offers more than just new forms of word symbols. More importantly, however, this language could be explored using computer animation, and could well lead to startling results.
The second is more fundamental. When we look at a work of architecture, we may be aware of some very subtle kind of perceptual relationship between ourselves and the building that is more than a simple visual appreciation. In some sense we may feel the weight of the stone, or the strength of a column. This kind of kinaesthetic awareness is to do with the ways that sensory information of one kind (visual) may somehow spill over into other kinds (tactile). Psychological research leads us to believe that human beings have a natural ability for these modes of perception, but that our present paradigms prevent us from developing them, so they become repressed. This could be an important area for exploration - the re-awakening of lost senses.
I think I have said enough to indicate that computer animation challenges us at quite fundamental levels. We return, therefore, to my first and last question : answer or problem? A catch question, of course - every answer creates a new problem, and as a general rule each new problem is bigger than its predecessor by an increasingly large factor; so now that we're close to the answer, perhaps it's time to look at the problem.