This article first appeared in the Journal of the Society of Almand Television Arts (Vol. I. No 2 2/3 1975) and is reproduced with kind permission of the Society.
The Antics computer animation system, first presented at the BKSTS Film '73 Conference, was developed by the author at Grove Park Studio and has been operating at the Atlas Computer Laboratory under the auspices of the Royal College of Art. It is currently the most sophisticated system of its kind in the world; the author here explains some of its capabilities and its limitations.
The appeal of animation as a medium lies in the fact that the designer has very great freedom to create whatever images the imagination desires. This freedom, in fact, is the main reason why animators are prepared to undertake such tortuous labours. So unless the computer can get at least close to matching this freedom, the majority of designers and animators are not going to be wholeheartedly interested. This is how I described one of the main aims of Antics when I first wrote about it in the December '73 BKSTS Journal. We are now at the point where we can say this aim has been achieved. One way of expressing the situation is to compare the capabilities of Antics with the work involved in a conventional film like Yellow Submarine. It would now be feasible to consider using Antics to make such a film; the comparison is not exact, because there are many things in Yellow Sub that Antics cannot do as well, and there are many things Antics can do better. However, most important of all is that it could be possible for a complete feature film of this kind to be made by a single animator working alone.
First though, let us put Antics in its context by describing quickly the overall state of computer animation techniques. Roughly, we can divide computer animation into three categories: (a) exploration, (b) simulation, (c) graphic animation. All films have something of all three qualities, so these are not isolated categories; however, computer animation systems generally place major emphasis on one category
The first, exploration, refers to the artistic exploration of film-makers like John Whitney and Lillian Schwartz. Abstract mathematical work of this kind is the easiest to do on a computer, so it is not surprising that most computer films have been of this kind; unfortunate, too, because many people have got the false assumption that computers can only produce abstract mathematical patterns. Artistic exploration is always a valid activity, but most film-makers are concerned with using animation as a means of communicating ideas and themes, and this involves exploiting and embellishing the existing languages of visual design, not in inventing new ones. Animators are quite right in regarding abstract exploration as a highly specialised form of the medium.
Simulation is the bread-and-butter work of computer animation: a proven and valuable tool that allows scientists and engineers to study complex processes in revealing new ways. The films produced are often nothing more than a convenient way of printing out the results of some enormous calculation; sometimes they are used for educational purposes, but even here the fact that every simulation has to have a special program written for it precludes the possibility of employing professional graphics techniques. Only fortuitously do these films have great visual impact. The exception is systems that aim to simulate the appearance of solid objects moving in perspective space. One such system is available as a commercial service to animators -Syntha Vision in New York. This will produce realisticlooking full colour animation of apparently solid objects, quickly, and at competitive cost. The main drawback is that it is a 3D simulation system, and will therefore not accept two-dimensional drawings. Everything has to be built up out of simple solid shapes - spheres, cones, cubes, etc. - with the result that it is tedious to work with, and everything produced tends to have a similar Toy Town quality. It would be more accurate to regard Syntha Vision as computerised puppet animation.
This leaves the third category, graphic animation - here we deal with systems whose main aim is to provide a useful tool for designers and animators. This area is the most neglected, since the animation industry is not large enough to fund research, and people in other professions do not feel it is any of their business. Consequently, we find the only three systems currently achieving genuinely useful results have all originated in obscure corners, far removed from the world of big organisations.
First, and best known, is Scanimate - operating in three studios in the USA, and one in London. As you may know, this is not strictly a computer in the ordinary sense, but rather a video manipulation device. It accepts artwork images fed in through a TV camera, and by manipulating knobs on a control panel this image can be made to distort, squeeze, bend, twist or whatever in a vast number of different ways. Because it is an analogue system controlled directly by the touch of a knob, Scanimate can handle sophisticated full-colour images with instant real-time playback something that digital computers cannot achieve at present. When the animator is satisfied with the playback, the sequence is finally transferred to videotape. The main disadvantage is that its animation capabilities are strictly limited - only five different colours can be used; it cannot do full figure animation, or moving backgrounds, or many other things that cel animation can do. So although it has been very successful in the field of TV graphics, it cannot be regarded as an all-purpose system.
Second is the system devised by the National Research Council of Canada, with the assistance of the internationally known animator, Peter Foldes. Using an extremely small and cheap computer they have made a system more or less tailored to Peter Foldes' style of animation - he uses line drawings, with animation largely consisting of smooth transformations from one drawing to another. Using the NRC computer, he has just made a film called La Faim, which was the official Canadian entry at the Cannes festival. It is a savagely powerful film, and deserves special place as a milestone in computer animation - most people seeing it unawares would not guess it had been made on a computer. Once and for all this film disproves many of the false assumptions that surround the idea of computer animation - La Faim certainly cannot be criticised as lacking passionate vision, nor of being inhumanly mechanical.
In designing the Antics system we have incorporated the animation principles of both Scanimate and of the NRC system; having more powerful equipment than NRC, however, we have now been able to go some steps further. Our major limitations are that being a digital computer system Antics cannot operate in real time like Scanimate, and there are also some limitations on image complexity that I shall describe later. The design philosophy of Antics is best seen in relation to conventional animation. Conventional work embraces a wide spectrum from the very simplest techniques like cut-out animation, where whole films can be made quickly and cheaply, to the most beautifully executed masterpieces of the animation-craftsman who has the patience to enjoy the labour of frame-by-frame drawings. In between these extremes is a large area of film-making where traditional assembly-line techniques are employed. Here there is a sharp distinction between the creative design work, and the work of producing the result. The distinction can be as sharp as it is between the architect who designs a house and the builder who builds it. Like the architect with his drawn plans and written specifications, the initial creative work of animation has two kinds of components: (a) drawings such as design drawings, character construction drawings, layouts, spacing guides and key drawings; (b) specifications in the form of bar sheets or dope sheets. The rest of the work is purely mechanical. The specifications describe how the design drawings are to be combined, what in-betweening is required; they describe how they are to be coloured and photographed, and they describe what standard operations (like pan or zoom) are to be used. Conventionally, this mechanical work is done by animation assistants, in-betweeners, tracers, painters and cameramen.
Essentially, Antics carries out all the mechanical work on the computer, while leaving the initial creative work as near as possible the same as in conventional animation. We begin, therefore, with drawings, which are fed directly into the computer using a device called a D-MAC pencil follower. This is like a large drawing table; you stick your drawing down and trace round it with a special kind of magnetic pen. The D-MAC senses the movements of the pen and automatically records the drawing. All drawings are fed in in the same way, whatever use we may wish to make of them. Besides using them as design images to be animated on the screen, they may be used as character construction skeletons; as grids to define how another image is to be distorted; as spacing guides to control the movement of another image; as backgrounds; as key drawings; or as graphs to control any other kind of change.
Once we have fed in the drawings, there now remains the specifications. These we chart out in a form resembling a cue sheet. Conventional animation charting deals with two things: a small vocabulary of key words like frame, field, pan, zoom, spin, fade, etc., plus some vital numbers - frame numbers, field sizes, start and end calibrations for a pan; in addition, drawings are numbered and assigned to numbered cel levels. Essentially, the Antics specifications are much the same, only the chart format is different.
There are 40 key words in the Antics system, each referring to one kind of animation operation. Some of them refer to basic displacement operations, mostly identical to camera movements: PAN, TILT, ZOOM, SPIN, HOLD, BANISH.
Others perform distortions similar to Scanimate, or the effects of techniques like wobble-glass: WAVE, SHAKE, WOBBLE, EXPAND, GROW, WIG-WAG, BENDY, SPACE, SHIFT. Other operations are complete animation systems in themselves: CHANGE will carry out smooth transformations from any image into any other, and on its own is able to produce animation like Peter Foldes' La Faim. SKELETON will carry out fullfigure animation; it will take a few simple matchstick skeleton drawings as key-frames; it will in-between these; it will then fit a character of any complexity on the skeletons frame-by-frame to produce smooth character animation. Furthermore, skeletons used to animate one character can be stored and used again to animate another character. SKELETON combined with CHANGE will allow the character to change from side view to front view (or whatever) at the same time as following the skeleton animation, so we can achieve a fully solid-looking effect with characters able to turn right round. SKELETON can also use grid drawings to produce freely selective distortion effects, or to achieve apparent solid effects such as a rotating globe of the world. FLIP and FLAP perform perspective rotations similar to the conventional flipover; TUMBLE and TURN perform perspective manoeuvres similar to a rollover. MASK is used to mask a drawing invisibly; SUPER allows complex matt effects to be achieved; MIRROR works like a mirror, and can be used to create kaleidoscopic effects; CYCLE is used to repeat sections of animation; PATH is used to make an image follow a freely draw movement curve; LEVELS is used to play about with the order of eel levels to achieve effects like having lots of images rotating round each other; FREAK introduces controlled random distortions into an image; CURLY will turn an image into an abstract geometrical pattern and play with it; TWIST will twist an image like a Christmas decoration and rotate it in apparent solid perspective.
There is no particular limit to how many of these things you can have happening all at once. You can have up to 130 different picture elements on the screen at once, all doing different things. Even on their own, most of these operations can be used in a variety of different ways, and even the simple ones have a flexibility that is difficult to compare with the conventional equivalent. For example, on a rostrum, you are lucky to get a zoom ratio better than 20 : 1; the Antics ZOOM will give a ratio of 10,000: 1. FLIPs can be combined and controlled to give the effect of a rotating solid cube. TUMBLE can be used to take an image, curl it up into a solid-looking cylinder with perhaps different colours on inside and outside, and maybe perforated with holes - and then set the whole thing turning, finally to return to its original flat state. Even the humble PAN can be used to take a complex image of a molecular structure and make it appear to rotate in three-dimensions. With suitable ingenuity, almost anything can be achieved.
However, this is only half of it. We command an operation to be performed by simply typing the numbers needed to describe it: for example, the start and end size of a zoom might be specified as 25 and 150 per cent respectively. However, apart from the 30 operations listed above, there are a further group of ten operations which are not applied directly to pictures, but only to numbers used in specifications; these ten operations are called CONTROLS and are similar to the concept of voltage control in music synthesisers like the Moog. The SINE control generates a sine-wave of specified period and amplitude: for example, varying between 25 and 150 with a period of 50 frames. If we apply this control to the start size of the zoom, then the image will continually zoom back and forth between these two sizes. Controls can be applied to almost every aspect of every operation - even CHANGE in-betweening can be controlled in this way, for example. Other controls are: LINEAR which produces a linear change from one value to another; WEDGE does a change from one value to another and back again with holds in between; WANDER takes a freely drawn curve as a timegraph of the desired quantity, and is particularly useful for lip sync animation; RANDOM produces controlled random fluctuations for animating things like flames; TWEENY in-betweens things in a curiously spasmodic manner that has many unexpected uses; CURVE controls things in a parabolic fashion, like falling objects; TAG hitches an image on to some part of another image; TAPER produces various cushioning effects; PHASES allows things to pass smoothly from one state to another, then another, then another, and so on in indefinite controlled succession. You can take it still further. You can use a control to control another control. Instead of having the period of a SINE control fixed at 50 frames, you could have this period varying as well...
In short there is no end to the possible variety and flexibility with which things can be made to happen. Once we have prepared a chart of the animation required, we type in the specifications following a fixed format, and then give the command to run the Antics program. For a typical 30 sec sequence of average complexity this will take about 5 min on the Atlas Lab's ICL J 906A computer. The result of this is a line-test, stored on magnetic tape, which we can view immediately on a TV screen. If we want to change anything, of course, this is simply done by correcting the specification and running the programme again. If, like Peter Foldes, we wish line drawings as a final result, then the work is now complete, and we can put the line-test tape on to the SC-4020 microfilm recorder and get it plotted on to film. However, if we want full colour results, we take our line test tape back to the 1906A computer and run a different Antics program - one that does the work of the conventional paint/trace department. This program takes the line test images, colouring in all the areas level by level. It also distinguishes between opaque colours and transparent colours, and if required can make a shape on one cel level merge invisibly (ELIDE) with a shape on another level; this allows shapes to penetrate and pass through each other. The program outputs each frame as a scanned image, similar to a TV image, with three separate black and white negatives corresponding to red, green and blue colour separations. Again, these are stored on magnetic tape, ready to be plotted on to film by the SC4020 microfilm recorder; thence the results are sent to Technicolor for triple-printing the finished full-colour film.
It is this last stage - plotting on to film - which constitutes the major snag in the Antics system. The Atlas Lab's SC020 plotter is, in computer terms, practically an antique, and is too slow and unreliable for Antics work. Symptoms of the SC4020 are: (a) slight raggedness is visible on edges of colour that are nearly vertical; (b) airbrush-type smooth shading is impossible; (c) plotting faults often result in spoiled negatives, which then have to be replotted all over again. This is a serious bottleneck in the whole process. Luckily, the SC4020 is due to be replaced by a new high-speed FR80 plotter which will overcome these problems. However, there is also a different kind of film-plotting machine being built at the Computer-Aided Design Centre in Cambridge. This converts the Antics scanned images into standard video format before recording on film. This will have all the advantages of the FR80, but will be much cheaper; it also opens up the possibility of working directly on to videotape, thus giving us real-time playback, like Scanimate. This brings us to consider some of the other possibilities for future Antics capabilities.
At present, we use key-frame matchstick skeletons to achieve full-figure character animations. However, we are currently investigating an even simpler possibility the use of Benesh notation. This is the music-like notation that choreographers use to script dance movements, and this could easily be adapted as a notation for Antics figure animation.
The Atlas Lab has an optical scanning device, which could be used like a kind of digital TV camera so that drawings - and ultimately even photographs - could be fed in directly. We have written a program for this, but not yet implemented it.
The animation program itself could be operated interactively on the Atlas Lab's PDP15 computer, allowing movements to be controlled directly by touch, in the manner of Scanimate, though only in line test form.
Finally, there is one Antics facility that I have not even mentioned yet. Besides drawing pictures, Antics also has a program that draws optical sound tracks. This in itself is quite a sophisticated thing. It operates in a manner almost identical to electronic synthesisers like the Moog or the VCS3, but with several important advantages. A synthesiser consists of a number of electronic gadgets that generate sounds such as sine waves, square waves, white noise, and such. In addition to these sources of sound, synthesisers also have various devices to modify sounds filters, reverberation units, ring-modulators and envelope shapers. Complex interconnections of selected gadgets can be discovered that will produce a vast variety of sounds - effects like wind, rain, storms, explosions, guns, squeaks, bells, drums, and fair imitations of most musical instruments, as well as musical sounds unlike any natural instrument. In synthesiser terminology, a specific arrangement of devices (corresponding to a specific kind of sound) is known as a patch. The Antics sound track program works in the same way. An Antics 'patch' consists of a specified arrangement of sources and treatments: the same devices are used in Antics as in a conventional synthesiser, so that it is possible to use a synthesiser to design a patch and then transfer it to Antics and get the same sound. Antics patches can be stored on punched cards and used over again; another advantage is this: on synthesisers the frequency of a note is set by turning a dial and it is not usually possible to set a second note to be exactly locked to a harmonic of another; in Antics patches this is possible, because frequencies can be specified as precise numbers, or as precise ratios of other frequencies. This allows complex sets of overtones and harmonics to be used. In the Antics program, you can have up to 20 different patches all producing different sounds simultaneously, so this is equivalent to having 20 synthesisers at your command. To generate music, the Antics program will accept a musical score; this may have up to 20 tracks, each with its own melody and its own patch. Notes are written simply as A, B, C, CS for C sharp, and so on, plus the number of beats duration of the note, and a digit to identify which octave the note is in. Consequently, with a selection of patches giving sounds of different instruments, and a track of music for each patch, it is possible to create an entire band.
The Antics sound track program will also accept hand-drawn shapes and use them either as wave-forms, or an envelope for other sounds. In effect, this is an automated version of Norman McLaren's hand-drawn sound tracks. The shapes produced on the sound track are often quite striking - our first musical test was Maxwell's Silver Hammer rendered in square waves; the resulting sound track included shapes that looked like little hammers. Of course, it is also possible to use the sound track shapes as picture - with different tracks in different colours- but this is something we have not yet had time to explore. The timing of sound tracks is measured in frames, which underlines the fact that the program is essentially intended as an accompaniment to the animation program, giving the animator the facilities of an electronic music studio, but able to produce sound tracks tightly synchronised to his animation. This brings us back to our starting point - the basic purpose of it all.
In an existential sense, anyone engaged on a mechanical task has become, for the time being, a machine. As a society we are beginning to move away from situations where one individual performs mechanical tasks for the benefit of another. Animation has always been severely handicapped by the amount of mechanical labour involved. It would be difficult to estimate how many films have been abandoned at the ideas stage simply because of the labour and cost that would have been needed to make them; it is also difficult to imagine what the animation industry would be like if everyone in it was directing their own films, rather than most people working on someone else's film. This indicates something of the untapped potential that exists for animation; if computer animation can release this potential it seems quite likely it would have a revolutionary impact on the whole area of visual communications.