The Future of the Cartoon Feature Film

Compute,

Model

or Draw?

A Research Report for the
MA in History of Film and Visual Media
David Mitchell
Summer 2002


Contents


  Preface
      Acknowledgements
      A Personal Note
  Introduction
  A Brief Survey of Animation Techniques
  An Overview of CGI Animation
  The Impact of CGI on Stop-Motion Animation
  The Impact of CGI on 2D Cel Animation
  Animation and Audiences
  Conclusions - Likely Futures
  Appendices
      Appendix 1 - Moore's Law, Hardware Technology and Computer Graphics
          Moore's Law
          Computer Graphics
          Motion Blur and Particle Systems
          Computer Games
      Appendix 2 - Box Office and Budget Figures
      Appendix 3- Biographical Notes
          Tim Burton
          John Lasseter
          Peter Lord
          Nick Park
          Henry Selick
      Appendix 4 - Forthcoming 3D CGI Films
  References
          Films
          Journal Articles and Books
          Radio Interviews by Max Schönherr for German National Public Radio
          Web Sites
          Newsgroups

Preface

Starting with Pixar's Toy Story in 1995, computer-based 3D animation has met with both critical and commercial success. This research project examines the current and likely future impact of this success on feature film animation, particularly on the more traditional, well-established techniques such as stop-motion 3D animation and 2D cel-based animation.

The primary source material for this research report has of necessity been the trade rather than the academic press. Such material, much of which is now published on the web, is of course influenced by the marketing strategies of those individuals and organisations whose products are being discussed. This is particularly the case where the article in question is an interview with the director of a film on the point of being released. This raises very real questions about the reliability and trustworthiness of any 'facts' appearing in such articles. Where possible I have tried to crosscheck controversial or surprising assertions and have looked at box-office performance as a more objective measure of the success or failure of a film.

Rather than fill the body of the report with lengthy details of the technology behind of CGI, I have placed most such material in an appendix. Other appendices contain additional supplementary material such as box office figures for recent CGI films, brief biographical notes on key animators and a list of forthcoming CGI features.

Comic strips and animated films are at least as popular in Japan as they are in the United States and there is a very strong animation industry in Japan. Its impact on the production methods of Western animation has been small however, so I have made relatively little reference to it.

Acknowledgements

I should particularly like to thank Wendy Jackson Hall (www.jacksonhall.com) for her willingness to answer my questions, Max Schönherr ( www.maxschoenherr.de) for helpful suggestions and permission to quote from his interviews conducted on German public radio, and Pamela Hutchinson for useful pointers to box office and sales statistics.

A Personal Note

Most of my professional career was spent designing and writing computer software. (1) My own knowledge and skills thus make it likely that I have occasionally used terms and concepts from computer science without properly defining them for the lay reader. I apologise in advance.


Introduction

The term CGI is a misnomer - the computer doesn't generate the images. That would be like calling traditional animation Pencil-Generated Imagery. No matter what the tool is, it requires an artist to create art.

- John Lasseter

Nine years ago Tim Burton's The Nightmare before Christmas (Selick, 1993) (2) impressed animators as well as film critics with its dazzling mastery of stop-motion animation. Coupled with the Oscar won that year by The Wrong Trousers (Park, 1993), it might have seemed that stop-motion, for years an animation technique largely confined to television advertisements (and art-house shorts from Eastern Europe) was at last becoming a serious rival to cel animation.

Just two years later Toy Story (Lasseter, 1995) appeared and the world changed. Computer-Generated Imagery (CGI) had already been used in live-action films (3) and John Lasseter (4) had been making CGI shorts at Pixar since 1985 (including Tin Toy (1988) which won an Oscar), but Toy Story was the first full-length film to be entirely computer generated.

While CGI slowly permeated the world of live-action films, it has had a far more dramatic, and rapid, effect on the animation industry. Indeed, it exhibits all the features of a 'disruptive technology'. (5) Thus, when the first ever Oscar for Best Animated Picture was awarded in 2002, it went, not to Disney, the studio that for sixty-five years has been synonymous with the cartoon feature film, but to a newcomer to animation, Dreamworks SKG, for Shrek (Adamson/Jenson, 2001). Disney themselves submitted Monsters Inc., a CGI cartoon produced by their partner Pixar, rather than their own traditional cartoon, Atlantis: The Lost Empire.

The success of CGI animation raises many interesting questions. Will traditional techniques, particularly stop-motion, be replaced by CGI? Will the increased realism of 3D cartoons make the form more popular, particularly with adults? Or is 3D CGI animation a 'flash in the pan' of which the public will quickly tire? What will be the effect of the close relationship between CGI animation and the technology used in computer games? Are there significant limits to what CGI can achieve and if so how will these limits affect the sort of animated films that get made?

The primary focus of this research report is on the first question - on the impact of CGI on animation techniques, but in answering that question we will have to look at some of the other issues. To understand the rapid growth of CGI, we need to appreciate the factors that have led to the current state of the art, so it will be necessary to examine the underlying trends involved in the development of computer hardware and graphics software. (6) To see why it has had such a powerful effect on animation, we will need to look the history of animation techniques as well as the economics of animation features. Since the technological and economic forces involved are likely to continue to exert an influence, this understanding will help us to predict the likely future.

CGI is a tool, and like all tools, it amplifies human powers in the performance of a task. This has two consequences. First, a tool presents an interface to a person (an HCI or Human Computer Interface). The features of this interface determine what it feels like to use the tool, which in turn affects not just how people use it, but what sort of people want to or like using it. Secondly a tool makes some tasks easy and some hard, which in turn affects the sort of products people make with it. To understand the impact of CGI on animation style we need to look at its HCI aspects and compare them with those of drawn and model animation.

It is easy to be misled by the marketing hype surrounding any new technology. While CGI animation has received a great deal of media attention, success at the box office is what determines how pervasive the technique becomes. (7) For animated features, box office success is not a simple matter of counting cinema admissions, since much of the revenue for such films come from video and DVD sales as well as from merchandise such as toys, books and video games.

A Brief Survey of Animation Techniques

Starewicz's influence is too great to measure, but The Mascot (1933) is without a doubt a precursor to the work of Henry Selick (Nightmare Before Christmas), John Lasseter (Toy Story), The Brothers Quay (Street of Crocodiles) and Nick Park (The Wrong Trousers, A Close Shave).

Eric Schneider (8)

Animation is older than cinema, indeed almost as old as photography. Niépce made the first still photograph in 1826, just six years before Plateau invented the Phenakistoscope. The Zoëtrope appeared just a year later in 1833. These two devices were examples of what today we would call drawn animation, presenting a series of pictures to the viewer in rapid succession to give the illusion of movement. While such toys were highly popular in the nineteenth century, it was not until the birth of cinema at the end of the century that animated films could tell stories.

Before the advent of CGI in the 1980s, animation techniques could be divided into two broad categories - drawn animation and model (or stop-motion) animation. (9) The former involves photographing a series of two-dimensional images, usually drawings but sometimes cut-out shapes, while the latter uses three-dimensional puppets and models. Both techniques developed rapidly in the early years of cinema, with Cohl (Fantasmagorie, 1908) and McCay (Little Nemo, 1911)among the drawn animation pioneers and Starewicz (The Revenge of a Kinematograph Cameraman, 1912) the pre-eminent puppet animator.

The invention of cel animation by Hurd in 1914 was a key milestone. Not only did it reduce the work required to produce drawn animation by eliminating the need to redraw the backgrounds, it also made it possible to divide the work up among a team of specialists. One artist could design the characters, (10) one draw the backgrounds, another produce key character frames as outline drawings, while yet others would work on the less inventive tasks of inking in outlines or filling in character movements ('in-betweeners'). The development of 'rotoscoping' by the Fleischer brothers was another key improvement, leading to much more realistic character movement.

While the cel technique transformed drawn animation into a streamlined production-line process, model animation remained very much an individual art. As a result it languished, while by the 1930s cel-animated shorts had become part of nearly every film programme. As well as Disney, several of the studios, Warner Brothers and MGM in particular, had flourishing animation departments. Disney produced the first full-length animated feature, Snow White and the Seven Dwarfs in 1937 and from then on released a new one roughly every other year.

The advent of widespread television in the 1960s had a profound effect on film programmes. (11) First newsreels were abandoned, as television news proved more efficient and immediate. Cartoon shorts were edged out more slowly, but became rare by the end of the 1960s. The animation studios, instead of making five-minute shorts for cinema exhibition, started making thirty-minute TV cartoons for children (Top Cat, Deputy Dawg andThe Flintstones for example). (12) While stories of necessity became more complex, the cartoon drawings themselves became highly simplified to meet the budgets that TV imposed.

Within the domain of animated shorts almost every conceivable technique has continued to be used, but until recently cartoon features have been produced exclusively using cel animation. Indeed, for more than forty years, from Snow White (1937) until The Secret of NIMH (Bluth/Goldman, 1982) almost all the feature-length cartoons came from a single studio, Disney. (13) Model animation, while often used for special effects in fantasy and science fiction films, (14) was not used for a feature-length animated film until Nightmare (1993).

Despite the success of Nightmare we have not seen many more stop-motion 3D cartoons. Indeed there has been just one, Chicken Run (Lord/Park, 2000), (15) although the long middle section of James and the Giant Peach (Selick, 1996) is entirely stop-motion. Instead there has been a series of very successful 3D cartoons generated completely by computers, starting with Toy Story in 1995. Before we consider whether there is a causal relationship between the apparent failure of stop-motion and the undoubted success of CGI, we need to investigate how and why the latter has developed.

An Overview of CGI Animation

Artists long for limitations; excessive freedom casts us into a vacuum. ... Style used to be an interaction between the human soul and tools that were limiting. In the digital era, it will have to come from the soul alone.

- Jaron Lanier (16)

CGI is good for making the fantastic seem real, whereas puppets are more suited for making the fantastic seem credible, which is not the same at all.

- Barry Purves (17)

The current state of CGI is the result of developments in two areas in computing - improvements in algorithms for computer graphics and the increasing density of integrated circuits. The software innovations have made it possible to generate and animate complex, photo-realistic scenes, while the improvement in hardware has meant that the calculations required to render each frame of such scenes now take minutes rather than years. One key stimulus for invention in the field is the annual SIGGRAPH conference, one of the largest computer conferences held each year, whose film programme acts as a showcase for the latest innovations. Throughout the second half of the 1980s, John Lasseter's shorts for Pixar (Luxo Jr., Red's Dream, Tin Toy, KnickKnack) were key highlights. Another equally important stimulus comes from the computer game industry, which has had a profound effect on computer graphics. Appendix 1 discusses the technology and trends behind the development of CGI in some detail.

Producing a CGI scene involves three basic steps. First a 'wire-frame' geometric model of every object in the scene must be developed. For animate objects that move, such models include the underlying skeleton and joints as well as the surfaces, since these define the 'avars' (articulation variables) that control the object's gait. As might be expected, special software tools have been developed to simplify this task. The second step involves defining the surface appearance of each object; its colour, texture, reflectivity, transparency and so on. Again there are specialised tools, known as 'shaders', to help. (18)

The final step is to 'render' the scene, that is to calculate the colour of all the pixels. As well as the models of all the objects in the scene, rendering also requires definitions of the camera and the light sources. Then the scene is ray-traced by drawing imaginary rays from the viewer's eye, through a window representing the screen, until each ray hits a point on an object within the scene. The colour of that point determines the colour of the pixel in the window through which the ray passes.

To determine the point's colour the illumination of the point is then calculated by drawing rays from the point to each light source, taking into account any other objects on the way. The absorption, opacity, transparency and reflectivity of each surface have to be taken into account. Note that this exactly reverses the process by which the viewer perceives the scene. Given the large number of pixels in a high-definition scene (several million) and the complex path a ray may take from light source to viewer's eye, this is an exceedingly complex task. Pixar's RenderMan is a good example of the sort of rendering software package required to do full computer animation. (19)

Approaching photo-realism has required many algorithmic innovations to cope with simulating partial reflection and refraction, shading and texturing. Work on fractals has made generating mountains, trees and other landscape features easier. Realistic skin, hair and fur is a recent achievement, as is realistic animal movement. It is not accidental that the first computer-animated full-length feature, Toy Story (Lasseter, 1995), had a cast made largely of plastic, wood and metal, with simple jointed limbs and a limited range of expressions. The dog and human beings in the film are the least convincing characters. Similar reasons lie behind the choice of insects, with rigid limbs and exoskeletons that look as though they are made of shiny plastic, in both A Bug's Life (1998) and Antz (Darnell/Johnson, 1998). The state of the art has advanced rapidly in the last few years, as seen in Shrek (Adamson/Jenson, 2001), Monsters Inc. (Docter, 2001) and Final Fantasy: The Spirits Within (Sakaguchi/Sakakibira, 2001).

Hardware power determines how long it takes to achieve rendering of a given level of realism. It is now possible for a home computer to do in real time 15 times or more a second (in games like Tomb Raider or Grand Theft Auto) what would have taken hours or days on a mainframe computer 30 years ago. However the level of realism in a film like Shrek is still a long way from being achievable in anything like real-time. Here are some statistics for Toy Story that reveal the state of the art in 1995: (20)

Minimum time to render a frame of film

45 minutes

Maximum time to render a frame of film

20 hours

Total frames

110,000

Size of completed film

500 GB

Size of Pixar's Render Farm (working 24 hours/day)

110 Sun workstations (21)

Toy Story Statistics

As can be seen, given the time it takes to render a frame and the large number of frames in a feature-length film, a whole battery of computers (a 'render farm') is required. Even so it can take more than a year to complete a film. (22)

It is not enough to use computer graphics to generate a static scene, objects in the scene have to move - they have to behave. Making computer-generated objects and characters behave in realistic ways poses several quite distinct problems:

Features like waves and ripples in liquids, flames and smoke, clouds and fog, have all become easier to do as computer power has increased. One difficult problem has been reproducing the flexing of complex surfaces such as the skin of the face or clothing. (23) Modelling a plastic toy such as Toy Story's Buzz Lightyear, whose facial expressions are largely limited to movements of gross features such as eyes, mouth and chin, is relatively simple. By contrast, trying to model a humanoid face such as Shrek's requires an extremely complex model of the underlying facial musculature. Even Woody's face had over 200 avars that an animator could control.

In 1987 Craig Reynolds, then at Symbolics, wrote a key paper (24) on modelling the behaviour of groups of animals, such as flocks of birds. It turns out that three simple rules, followed independently by each bird, lead to complex flocking behaviour, an example of what is called an 'emergent' property. The distributed algorithms he described have since been used in films such as Batman Returns (1992) to model bat swarms and penguin flocks and Jurassic Park (1993) to model stampeding dinosaurs.

So far we have concentrated on the use of computers in 3D cartoons. It is worth pointing out that computers have long been used in conventional 2D animation as well. Within the production-line approach of cel animation, computer graphic tools were first used to assist 'in-betweeners', with the development of so-called 'tweening stations'. While these at first did little more than simple 2D morphing between one key frame and the next, the process now involves a 2D analogue of the wire-frame model used in 3D animation, complete with underlying (flattened) skeleton and articulation points.

While final rendering takes a great deal of computer time, it is perfectly possible to view real-time animations during the design phase, not only as simple wire-frame models, but as simplified versions of the final product. This makes it quite easy to experiment with alternatives. For example, the computer model incorporates details of the position of lights and cameras as well as the objects, characters and backgrounds in the scene, so the former can be changed without any need to redraw characters or backgrounds. This leads to one of the key differences between CGI and conventional animation - the ability to move the camera in an extremely dynamic fashion - something that is particularly notable in Pixar's films. (25)

While camera movement is perfectly possible in stop-motion and cel animation, both lead to problems. In the case of stop-motion, the movement has to be pre-planned since there is no going back without having to reposition the models and reshoot every frame from new angles. Camera movement in stop-motion is thus unlikely to be the result of spontaneous experimentation. Cel animation is faced with a different problem. As the camera moves, backgrounds change. Simple linear movement - a tracking shot - can be accomplished relatively easily without redrawing, but panning changes the perspective and can force redrawing of the whole background for each frame, a hopelessly uneconomic task.

Though it may take a great deal of computer rendering time, it requires little human effort to 'reshoot' a scene with repositioned cameras, something that is unthinkable for traditional animation. As well as allowing experimentation during production, this has other uses. When A Bug's Life was being transferred to video, rather than use the traditional 'pan and scan' technique normally used to convert from a wide-screen to TV format, the film was recomposed. Leo Hourvitz, Technical Director at Pixar, explains:

The original frame was carefully composed to the widescreen "camera", and where that composition was just wrong for the full frame, we went in and recomposed the painting, as it were. .... Because we're in a virtual environment, there is something out there. So in a lot of those shots, you can actually see more top and bottom than you did before ... in a lot of cases, you actually see more of the virtual environment than you did in the theater. (26)

There are two other specific capabilities that differentiate CGI from stop-motion - motion blur and particle systems - that are worth examining in some detail. For CGI motion blur is a straightforward rendering task, and it appears in most shots containing fast moving characters or objects. (27) By contrast, while motion blur is not impossible to accomplish in stop-motion, it is very difficult. It is only in rare situations that a model can be moved smoothly and sufficiently fast during the brief exposure of a single frame so that blurring occurs. (28)

Even when models can be moved, the price of motion blur is normally a static camera position. A recent SIGGRAPH paper by Brostow and Essa showed how motion blur could be added to conventional stop-motion sequences in post-production, by using a computer to analyse the differences between frames, detect fast moving objects and then blur them. (29) Note that this approach works even when the camera is moving - 'stationary' objects in the background will be (properly) blurred. As yet the technique has not been used commercially.

For many stop-motion animators though, the lack of motion blur is an asset not a liability; it adds to the unreal quality of the animation. As Ray Harryhausen says:

It's the dreamlike quality of it [stop-motion animation]. That's what makes it fantasy. Computer pseudo-realism doesn't give that. (30)

Not all objects in a CGI scene are rendered using standard ray tracing techniques. Particle systems are used to model things like fire, smoke, fireworks and the leaves of trees. Such features consist of a very large number of objects, all of which behave in a similar way. Ray tracing collections of many thousand such objects would require enormous computer power (even by today's standards), so special techniques have been developed to deal with them. (31) The motions of the individual 'particles' in such a system are typically quite complex, since they are often acting under various constraints. For example, the behaviour of the leaves of a tree swaying in the wind are determined by the wind (which may be gusting), the elasticity of the branches of the tree and gravity, the detailed, individual way they are attached to branches and the movement of those branches themselves. While the computations involved may be complex, computers now make it possible to produce realistic animated images of such (32) scenes.

Conventional stop-motion, unable to cope with the sheer number of individual objects, let alone deal with their complex motions, either avoids such scenes, or handles them using mattes and conventionally filmed live-action inserts. (33)

The Impact of CGI on Stop-Motion Animation

Model animation, by its very process, has a slight unpredictability and spontaneous feel to it - even the animators cannot exactly predict where the puppet will go - and this does give it a unique edge.

- Barry Purves (34)

I don't think after Jurassic Park that we can, or should, ever accept a model-animated dinosaur again.

- Barry Purves (35)

It should be obvious that although the success of 3D CGI cartoons is likely to affect all conventional forms of animation, it is with stop-motion that it competes most directly, and hence will impact most. After all, it is already possible, using CGI, to reproduce the vast majority of conventional stop-motion sequences. Further, Moore's Law continues to make CGI cheaper, while stop-motion costs are likely to increase. (36)

Things are rarely simple however. As was mentioned in the introduction, the tools an artist or craftsman uses can have a profound effect on their way of working. How do the undoubted differences between the techniques affect the end results? And do they affect the imagination and ideas of the people involved?

There is quite a lot of anecdotal evidence that stop-motion exerts particularly strong effects, both psychological and physical, on its practitioners. Adherents feel that the limitations and idiosyncrasies of the technique lend a very particular character to the end-result. Following a panel discussion of stop-motion animators, Stephen Arthur made these remarks:

Barry Purves inspired the audience with his impassioned descriptions of the sensual art of "directing" his elaborate puppets. He let us handle these expensive puppets: the original alien from Mars Attacks, the statue-like figure of Achilles, and his "14-inch Willy" (Shakespeare) from Next. It was an astonishing feeling, a very direct connection with the "actors" in puppet animation. Puppets are usually held intimately by the animator's hand for every increment -- a strong contrast to the separation from the animated flow experienced with most types of animation. I could suddenly understand what entices so many animators into puppet animation. (37)

Purves himself goes even further:

Because of the one-to-one relationship between an animator and puppet in the actual process of animation, a lot of the character and the passion of the animator himself goes straight into the puppet and its performance. It is the most purely personal and honest form of animation (often revealing surprising aspects of the animator's personality), and acting is the most undiluted, whereas with drawn and computer animation, it is, to some extent, animation by committee. There are, of necessity, so many more people involved, and something really does get lost on the way. (38)

There is a close relationship between acting and animation, indeed it has been said that animators are often frustrated actors who are too frightened to act in front of the camera, so they act with a pencil or a puppet instead. Because successful animation requires a deep understanding of how and why animals move, animators often take acting classes to develop their understanding of the relationship between mind and body. (39) In stop-motion, the direct physical manipulation of the puppet makes animation much more of a performance. As Peter Lord says:

[Plasticine animation] is a technique that's perfect for improvisation.... You have your camera, you have your puppet, and your animator knows the intention of the shot, but there are so many ways you can carry it out. So it's like any other performance. You may know you've got to keep the shot within three seconds, and it may take you a whole day to do, say, a second and a half, but you have so many opportunities throughout that day to change your mind on how the shot goes. And ideas come to you, because you've got all that time. With drawn animation, it's like you're creating a more sophisticated flip-book, so you've got to look through your drawings and keep going back and adjusting. And with computers, you can keep changing the animation and layering it. .... Here, you've got a puppet and you're just working forward, and if it doesn't quite deliver, you start again from the beginning. You really live in the moment, like an actor would on a stage. Each day is a singular performance. (40)

Supporters of stop-motion argue that CGI technology, where the puppet is a virtual figure, seen on a computer screen and manipulated though a mouse or keyboard, lacks this direct, tactile approach, and hence cannot duplicate the particular qualities of stop-motion. This need not be the case, as the following case study shows.

The Tyrannosaur in Jurassic Park (Spielberg, 1993) was originally intended to be filmed using a combination of a large 'animatronic' robot (built by Stan Winston) and conventional stop-motion models (animated by Phil Tippet). CGI was to be used purely for the stampede scene, which involved too many animals to be filmed using models. The early attempts to animate a Tyrannosaur using CGI were so impressive that Spielberg abandoned the idea of using stop-motion, although he did retain Tippet as a 'dinosaur supervisor'. Tippet organised classes in mime and field trips to zoos for the computer animators, but he also invented an ingenious computer input device, the Dinosaur Input Device or DID. He equipped the joints of a conventional ball-and-socket stop-motion armature with encoders that allowed any movement of the armature to be communicated to a computer. In this way a CGI model of a dinosaur could be manipulated using the techniques of traditional stop-motion. (41)

The DID was short-lived - by the time work on the film was completed the computer animators had switched back to using mouse and keyboard - but it does serve to demonstrate that CGI need not be an entirely 'virtual' activity. CGI Animators can physically touch and feel their creations through such exotic haptic devices. (42)

In any case, even when using a mouse and keyboard, aspects of the animator's personality still get transferred to the screen character, as Andy Jones, animation supervisor on Final Fantasy: The Spirits Within (Sakaguchi/Sakakibira, 2001), explains:

Roy Sato, for example, was the lead animator on Aki [Dr. Aki Ross, one of the leading characters in the film] and you see a lot of Roy in that character. The timing of Aki's blinks are the same as his .... (43)

The most obvious demonstration of the serious impact that CGI can have on stop-motion animation lies in the contrast between Nightmare and Tim Burton's subsequent film, Mars Attacks! (1996). Though Nightmare was a stop-motion triumph, its box-office performance was not outstanding - it cost around $23 million and made just $57 million. (44) Here are some statistics that may indicate how the money was spent:

13 principle animators
100 model makers and set builders
230 sets - often 20 or 30 feet square- on 19 stages
60 characters (more than 200 puppets)
400 heads just for Jack Skellington
1 minute of film completed each week
400 frames in a typical shot (which took 10 days to shoot)

Burton originally intended the Martians inMars Attacks! to be animated by Henry Selick using the same stop-motion techniques they had employed on The Nightmare Before Christmas. (45) Selick's company, Skellington Productions, was however still busy on James and the Giant Peach (1996) so Burton turned to two English model makers, Ian Mackinnon and Peter Saunders, whose credits included the creation of puppets for the Oscar-winning stop-motion short The Sandman (Berry, 1991).

Mackinnon and Saunders organized a large team of sculptors in Los Angeles and the U.K., building hundreds of identical 15-inch Martian puppets. Mackinnon was soon joined by puppet animator Barry Purves, creator of the award-winning short films Next and Achilles. With Purves acting as animation director, elaborate sets were constructed and filming began. After months of work designing Martian gestures and ways of moving, Warner decided that blending stop-motion animation convincingly with live-action was too challenging a task to be dealt with in the year left before the film's scheduled release. The model animation team was dispensed with, to be replaced by CGI animators from Industrial Light and Magic, although not all of the model work was discarded. The movements and gestures developed by Purves' team were adapted to the computer characters, the puppets were digitally scanned and rendered into computer models, and scaled up to build full-scale Martians for several of the film's live-action scenes.

Mars Attacks! was an expensive film to make, and not just because of the money spent on discarded stop-motion. Unlike Nightmare, which had no expensive stars, Mars Attacks! had a large cast which included major stars such as Jack Nicholson, Pierce Brosnan and Glenn Close. It also featured several large and expensive set pieces. Eventually costing over $70 million, it failed to recoup its costs, with a US box office gross of less than $40 million. As far as the public, critics and industry observers were concerned however, the failure of the film was not due to the use of CGI, but in spite of it - most felt the special effects were outstanding.

As has been mentioned, Henry Selick was working on James and the Giant Peach (1966) while Mars Attacks was in production. Selick's plan was to shoot the central character as a live action boy and have him interact with stop-motion creations through the entire story. When that proved too expensive he explored making every element stop-motion. Disney felt the cost was too high, so a compromise was reached. Of the result Selick has said:

The compromise was hard on a lot of audiences. I believe it would have worked better either way, all stop-motion or as James all live throughout. (46)

Before James was released, it was widely felt at Disney that it would do better than Nightmare. (47) As can be seen from the figures in Appendix 2, it actually did much worse. The abandoning of stop-motion on Mars Attacks! and the unexpected box-office failure of James convinced many in the industry that CGI was the way of the future, although Selick was still optimistic. However the collapse of a three-picture contract with Miramax in 1997 led him to close his studio in San Francisco and work on smaller, personal projects. The dismal box office performance of his latest film, Monkeybone (2000), which, like James, mixed live action and stop-motion, has discouraged most American studios from becoming involved with stop-action features.

The exception is Dreamworks SKG, which financed the very successful Chicken Run (Lord/Park, 2000) as part of a 250 million dollar, five picture contract with Aardman Animation. It is worth asking what is the secret of Aardman's success and whether they will be able to continue making feature-length 3D animated films using stop-motion when everyone else has switched to CGI. (48)

Part of the answer surely lies in Aardman's success with its Wallace & Gromit series. These three half-hour films have been extremely popular on television all over the world and, along with Creature Comforts, account for Nick Park's three Oscars. While the income from these short films has been minute in comparison with that from feature film successes like Toy Story or Chicken Run, they have created a very large, world-wide audience that is familiar with, and attracted to, the Aardman style. As Peter Lord says:

We've kind of developed this style with the eyes right together in the middle and a big wide mouth, just because that plays well. People find that incredibly funny for some reason. Just the still image makes them laugh, so we have to go along with that. (49)

As well as strong stories, the emphasis on humour is something that Aardman shares with Pixar. Both avoid the excesses of sentimentality characteristic of much of Disney (as well as the saccharine songs) and the dark expressionism that characterises much of Selick's work. Both share a delight in visual and verbal puns, in the spoofing of film genres and in rich visual detail. Indeed, there is a strong affinity between the 'look and feel' of Aardman's plasticine and that of Pixar's 'virtual plastic' (it is not just the toys in Toy Story that appear to be made from plastic, the insects in A Bug's Life do too). It is not surprising that both companies have a profitable sideline in merchandise - toys, T-shirts and mousepads, as well as books, videos and DVDs - based on key characters such as Wallace, Gromit, Woody and Buzz.

Successful as Chicken Run was, something may have been lost in the transition from shorts to features. The large number of characters involved meant that puppet making became more of a production line, with the use of moulded plastic instead of plasticine for some chicken body parts. With so many people working on the film (upwards of 200), keeping the animation consistent inevitably resulted in a lessening of the stamp of individuality that characterised the earlier, shorter pieces. As Nick Park recalled:

I remember starting A Close Shave. There were more animators than ever working on that one, and they were all put through Wallace & Gromit lessons, where they learned how to move the brows and the eyes and the mouths. We did that even more on Chicken Run. Every Monday night, we had workshops so everyone would handle the characters in the same way. (50)

Aardman's next feature for Dreamworks will be The Great Vegetable Plot, featuring Wallace and Gromit, and is due for release in the autumn of 2004. The company is also busy working on a variety of shorts, including a new series of Creature Comforts for ITV and a series of 1-minute Wallace & Gromit episodes to be released via the Internet.

Not everything at Aardman has been going as planned. The second Dreamworks feature was to have been The Tortoise and the Hare, based on Aesop's fable. Originally scheduled for release this summer (2002), and then in 2003, it now looks unlikely to appear before 2005. (51) In July 2001 Aardman announced that it had laid off 90 of the 170-strong crew because of script problems on the film. The Hollywood Reporter announced on June 11, 2002 that the film had 'failed to cross the finish line after script troubles', implying the project had been cancelled. Then on July 24, 2002, a spokesman for Aardman Animations revealed that work was continuing on the script for the film, which is now expected to have CGI animation augmenting its stop-motion work.

Despite their commitment to stop-motion, Aardman have been experimenting with CGI, though Peter Lord has said that trying to copy clay animation exactly using CGI would be a 'very sterile exercise because it is just copying' - what interests him is 'devising a new language' for CGI animation. As he says:

Well, there is something about working with the materials. There is a fundamental difference between working with your hands and your arms and your fingertips, and working on the keyboard. .... You grab the puppet with two hands, and you feel the whole thing move, you feel the twist of the chest away from the hips, the roll of the shoulders. . . . (52)

If stop-motion animation has had problems, disappointments and failures, so has CGI animation. As well as the disappointment of Mars Attacks! (which had a foot in both camps), there has been one CGI failure of truly epic proportions. Final Fantasy: The Spirits Within (Sakaguchi/Sakakibira, 2001) had a budget of $137 million dollars, but has so far recouped just $32 million dollars.

Final Fantasy is perhaps the most ambitious CGI feature to date. The film, which took four years to make, put synthetic human actors into roles that could easily have been played by real humans and placed them in completely synthetic sets. (53) The film, based on a series of hugely popular, interactive, role-playing computer games, was produced by Square, the company that produces the games, and co-directed by Sakaguchi, the game's originator. With world-wide sales of the nine-part game series totalling more than 26 million units, Square must have thought it had a ready-made potential audience of game-players familiar with the fantasy themes, comfortable with computer graphics characters, and eager to see the next instalment.

The CGI animation has been rightly regarded as a technical triumph, so why was Final Fantasy such a commercial failure? The answer, at least as far as western audiences and critics were concerned, (54) was the weakness of the story and the lack of pace in the way it was told. Lacking a compelling plot, the glamour that human stars can bring to a film, and without much humour, the film had little but its special effects to hold the attention of the audience. As a result of its failure at the box office, Square announced in February 2002 that it was closing the studio in Hawaii that had created the film.

Although Disney has relied on its partner, Pixar, to produce CGI cartoon features that it then distributes, it has made one attempt at a (partial) CGI feature itself, Dinosaur (2000). Although the dinosaurs in Dinosaur were created using CGI, the scenery was live action, with backgrounds shot all around the world. The film made $350 million at the box-office, but it cost $200 million to make - at one time nearly 900 people were working on it. Its poor financial performance caused Disney to close the CGI unit it had created to produce the film, though it has since re-opened it on a much smaller scale. (55) The accepted explanation for the failure of Dinosaur is the familiar one - a poor plot - which recycles ideas from The Lion King and Tarzan.

The Impact of CGI on 2D Cel Animation

We proceeded as all artists did before us: with pencil and paper. If anybody wants to be an animator, they should learn to draw the human figure.

- Chuck Jones (56)

If CGI competes head-on with stop-motion, it also competes with 2D cel animation. Largely aimed at the same audience, they are thus competing for the budgets that studios are prepared to spend reaching that audience. There is also a more subtle form of competition that occurs within the animation community itself, where a relatively fixed number of practitioners have to choose which tools they will use to realise their ideas. There is considerable evidence of a growing drift of animators towards CGI. (57)

As well as competing, CGI has also been used with cel animation in a co-operative manner. In the late 1930s, Disney developed the 'multiplane' camera system, an elaborate animation stand that allowed several separated cel layers (foreground characters and background sets), to be moved independently frame by frame, giving a powerful illusion of three-dimensional space. An updated version of this technique uses CGI to replace the background layers.

Films produced using this approach are sometimes called 2D/3D hybrids. In such films the foreground characters are handled in a conventional 2D manner (though often using computers to assist the animators to do tweening) but background scenery - buildings and trees for example as well as features such as crowds - are modelled using 3D CGI. This is becoming an increasingly popular way of producing 'traditional style' cartoons - Disney used CGI for backgrounds and crowd scenes on Mulan (1998) for example. Several of the forthcoming features mentioned in Appendix 4 are of this type.

One advantage of this approach (which it shares with 3D CGI) is that it makes it simple to adjust the position of the camera and even move it during a shot. While the 'multiplane' system allowed the camera to perform tracking and zooming shots, it could not cope with a true pan because rotating a constant background image introduces unacceptable perspective distortion. CGI offers much more freedom because the background is redrawn in the proper perspective for each frame.

Evidence for the impact of CGI on 2D cel animation is not hard to find. The success of the four features Pixar has so far produced (Toy Story (1995), A Bug's Life (1998), Toy Story 2 (1999), and Monster's Inc. (2001)) has not been matched by Disney's own traditional 2D offerings. The following table shows the budgets and world-wide gross box-office figures for the four Pixar features.

Title

Date

Budget

World-wide Box Office

Toy Story

1995

$30 million

$356 million

A Bug's Life

1998

$30 million

$358 million

Toy Story 2

1999

$90 million

$486 million

Monster's Inc.

2001

$115 million

$523 million

Pixar 3D CGI Animated Features since 1995

The second table shows the budgets and world-wide gross box-office figures for the seven Disney animated feature released since Toy Story appeared in 1995. (58)

Title

Date

Budget

World-wide Box Office

The Hunchback of Notre Dame

1996

$70 million

$303 million

Hercules

1997

$70 million

$250 million

Mulan

1998

$70 million

$303 million

Tarzan

1999

$150 million

$435 million

The Emperor's New Groove

2000

$80 million

$160 million

Atlantis: The Lost Empire

2001

$90 million

$139 million

Lilo and Stitch

2002

$80 million

$190 million

Disney 2D Animated Features since 1995

It is dangerous to read too much into these tables, but in contrast to the Pixar figures, there has been a clear reduction in gross box office takings for Disney's traditional offerings over the last few years. (59) The apparent success of Tarzan is countered by its very high budget, which in turn was due to the very large number of artists (over 570 at the peak) who worked on the film. It was from this point that Disney started making dramatic cuts in its overheads.

In March 2002 the Wall Street Journal reported that the company would lay off an additional 250 animators over the course of the year. When Lilo and Stitch was released in June 2002, Thomas Schumacher, president of Walt Disney Animation, said that Tarzan would have been a considerably greater financial success if it been made with the lower salaries and cost-control measures now in place. Tarzan generated an internal rate of return of 14% on Disney's investment but using the new processes the return would have been 35%.

Dreamworks SKG, which commissioned and distributed Antz (1998) and Shrek (2001), the two very successful CGI animated features produced by PDI, has also made traditional animated features itself in the last few years. As with Disney, these traditional features have been failing to match the box office success of the CGI releases, as the following tables show.

Title

Date

Budget

World-wide Box Office

Antz

1998

$60 million

$152 million

Shrek

2001

$50 million

$482 million

Dreamworks SKG (PDI) CGI Animated Features

Title

Date

Budget

World-wide Box Office

The Prince of Egypt

1998

$60 million

$221 million

The Road to Eldorado

2000

$95 million

$75 million

Spirit: Stallion of the Cimarron

2002

$48 million (60)

$97 million

Dreamworks SKG 2D Animated Features

Again it is dangerous to draw too many inferences, but there is a pattern here too. Note that both Road to Eldorado and Spirit are 2D/3D hybrids, using CGI for backgrounds and crowd scenes.

Animation and Audiences

The modern film tries too hard to be real. Its techniques of illusion are so perfect that it requires no contribution from the audience but a mouthful of popcorn.

- Raymond Chandler.

Most box-office failures are the result of self-inflicted wounds. It is rare for a film to fail because of external forces, unfair critical notices say, or a vengeful William Randolph Hearst using his power to suppress publicity for Citizen Kane. But as well as making mistakes in production - poor casting, plot failures, careless editing and so on - studios can make mistakes in marketing. Thus in August 1999, The Iron Giant opened in the US to excellent reviews. The film was hailed as a 'solid and ironclad winner', 'intelligent', 'well-crafted', and 'visually appealing'. With a production budget of $48 million, the film took only $23 million in the US. The trade press blamed Warner's poor marketing campaign for the box-office failure of the film. To understand such failures we need to look at the audiences at which such marketing campaigns are aimed.

Feature-length cartoons, that is, animated fantasies, have been with us for almost sixty years. For most of that time they were directed at the same primary target audience - families with young children. (61) This is reflected in the stories, which were often based on traditional fairy tales (Cinderella, Snow White) or children's books (The Jungle Book,Winnie the Pooh) and in the visual style, which matches children's book illustration.

More recent attempts to reach other audiences have rarely been successful. For example, Bluth and Goldman made Titan A.E. (2000) for Fox. The film, which had a budget of $85 million, made only $30 million at the US box office. Looking back at the film two years after its release, Don Bluth candidly confirmed:

Fox wanted the film targeted to 14-15 year old males. We felt that animation for this age/gender group could spell disaster. At the audience reaction screenings, prior to release, there were some very interesting arguments, among the members of the test audiences. The over 25's thought the skew was for under 17's and for males. The under 20's thought it was an older skew, targeted to both males and females. Parents thought it was too violent for their children. We think we hit the target that Fox asked for. We didn't agree with this target. Animation, traditional animation, seems to have a split audience, 6 to 11 and 19 to ?. We think that the 12 to 18 year-old audience abandons animation in favour of PG13 and any other adult product they can get into. They regard animation as too young, a product that they go to see with their parents and now they want to break free and be an adult, at least with regard to films. Titan was definitely a departure from the main stream animated product. It really needed promotion and advertising to fit the product target. It was a big mistake to advertise on Nickelodeon for a 14 year-old male-targeted film. (62)

In the wake of the failure of Titan, Fox closed the Bluth/Goldman studio.

Other studios have tried and failed to get to grips with niche audiences. Disney targeted Atlantis at 14 year old boys and Dreamworks saw Spirit as a vehicle to capitalise on the interest young girls have for horses. Neither film was particularly successful, but if studios have had difficulty attracting the teenage audience into cinemas to view traditional animated features, TV has had some success, as both The Simpsons and South Park testify. In fact, the recent South Park cinema feature, South Park: Bigger, Longer & Uncut (Parker, 1999) did reasonably well; with a budget of $21 million it grossed over $52 million at the US box office. It was rated R in the USA, and 15, 16 or 18 in most European countries.

Understanding the audience for animation is not easy and there is very little in the way of 'film theory' to help us. (63) There are three distinct forms of popular animation - the five-minute cartoon, the thirty-minute TV series and the full-length cinema feature - although only the last two are still being made. The TV series have much in common with TV situation comedies, indeed it is not unreasonable to see such cartoons as 'sitcoms for kids'. (64) We see a strong emphasis on broad 'physical' comedy, on verbal wit and word play and, perhaps most obvious of all, the signs of very limited budgets. (65) In the UK, Sky transmits seven channels almost entirely devoted to the genre and cable TV in the USA does much the same, so there is clearly a sizeable young audience.

While most of these cartoon series are clearly aimed at young children, there are some notable exceptions. The Simpsons and South Park have already been mentioned. It is worth examining both in more detail. Asked his opinion of The Simpsons, Chuck Jones said:

Well, I have a lot of respect for The Simpsons. They're very clever scripts, and they had no intention of animating them. Last night, when I was signing some cels, this deaf girl came up. She could read my lips, and she said that the thing she likes about the Warner cartoons and the Disney cartoons is that she could tell what was happening without hearing the dialogue. And that's what we tried to do: We always ran the pictures without dialogue, so we could see whether the action of the body would somehow convey what we were talking about. And she said that she'd watch The Simpsons and she couldn't tell what was happening, because so much of it is vocal. It's what I call "illustrated radio." The thing has to tell the whole story in words before you put drawings in front of it. (66)

Jones makes a good point. While The Simpsons has a clearly identifiable visual style, much of the humour is verbal - highly complex and referential with considerable sexual innuendo. The visual humour too is full of allusions - to famous films, paintings and books. While it does appeal to children, much of the humour is aimed well over the heads of six to eleven year olds. The humour of South Park - anarchic, indeed scatological - is even more clearly aimed at a teenage audience. The animation appears to be much cruder than The Simpsons, though this is the result of a conscious stylistic decision rather than mere cost cutting. (67) Wendy Jackson-Hall remarks:

South Park animators work surprisingly hard to get that effect - of bad animation. I knew of an animator on that show who was often sent back to re-do scenes she had animated too fluidly. (68)

Not all TV cartoon series use 2D cel animation. Max Steel (which only ran for three seasons) was produced entirely using CGI, as is the new British TV series, Dan Dare, Pilot of the Future. Both feature the same sort of synthetic human characters (or 'synthespians') as Final Fantasy, albeit of a more comic-strip character. As yet none of these 3D CGI-produced series has been really successful. It seems probable that the TV audience for thirty-minute cartoon series is more impressed by humour and wit and an idiosyncratic visual style than it is by 3D 'realism'. In Treehouse of Horror VI (1995), a Simpsons' Halloween special, a PDI-generated Homer3 gets trapped in a 3D CGI world. There has been little or no demand from the audience for a repeat of the experiment.

That experiment does however shed an interesting light on the psychology of perception. Homer in 3D should look more realistic than his flat 2D counterpart and in a simple sense he does. But he doesn't seem any more credible; indeed if anything he seems less believable, more artificial. Homer's 'reality' is a complex matter. Even in 3D he is clearly a fantasy figure - no living being could actually look like that. His existence, like that of Mickey Mouse, Dumbo or South Park's Cartman is entirely unrelated to the everyday world we inhabit. Such figures exist only in the fantasy universes of our imagination. (69)

I think this helps us understand the appeal of Toy Story. In a recent interview, John Lasseter said:

Ever since I worked with computer animation, there have been all sorts of people that have always desired and thought that [photo-realism] is the goal. For me, the way that we work is that we use sort of realistic imagery only as something to shoot for. We say that reality is just a convenient measure of complexity. Because if you can create a tool that can produce something that looks almost real then we like to take a step back and produce something the audience knows does not exist, that it's a cartoon, it's caricatured, it's fantasy, it's something, but then use these tools to make that look so believable in the world that we're creating. (70)

The creators of Shrek clearly agree. The original animated model of Princess Fiona was very realistic - she looked human. But next to an ogre and a talking donkey, a realistic-looking princess seemed out of place. At an early screening of the film, Jeffrey Katzenberg said:

When we had placed Fiona in the movie, which is a fairy-tale world, it looked completely wrong.

The co-director ofShrek, Andrew Adamson added:

With a talking donkey, you've got leeway because no one's ever seen one. Fiona had to be a little bit stylised so she fit into this somewhat surreal, illustrative world.

If the CGI cartoons we have seen so far have been fantasies, their success has surely come from the fact that they are fantasies designed to appeal not just to children, but to families. As Lasseter says:

Now will Pixar do a film for adults? We already have - we've done four of them, they happen to include kids too. And that's something I believe in very, very strongly. We can make a film that is fantastic for adults, is truly entertaining for teenagers, adults without kids as well as families. .... I think the work at Pixar fulfils a need in the world for this type of film. I also go to family films really just for kids and I'm bored silly, and I don't want to go back a second time, even if the kids do. I love the idea that adults love our films as much as kids do. (71)

Conclusions - Likely Futures

Future, n. That period of time in which our affairs prosper, our friends are true and our happiness is assured.

- Ambrose Bierce

We face an uncertain future, not least because of the rapid pace of technological change, so any attempt at prediction is likely to fail. Even so, some signs are relatively clear.

Computers seem guaranteed to get faster, in accordance with Moore's Law, for the next ten years at least. In itself this would just mean that rendering times would decrease, but we can also expect that CGI software will become both more powerful and easier to use. (72) More powerful, in the sense that it will be able to produce an ever wider range of effects; and easier to use, in the sense that the level of computer knowledge required will decrease. These trends could lead to significant changes. Thus Mark Sylvester (co-founder of Wavefront) predicts:

The barrier to entry has gotten so low that it is feasible to create a studio that is under the radar of traditional Hollywood - and therein lies one of the secrets to their success - low overhead, reasonably paid animators, and a passion to deliver high-end product, for low end prices. As more companies begin to specialise, it is not out of the realm of reason to see shops offering Cloth, Hair and Dynamics effects that cannot be completed anywhere else. (73)

When asked what was likely to be the next 'big thing' for the animation industry he replied:

Directed Characters, models with intelligence - behavioural control of animation, is my first answer. The real answer is making this technology that we have developed over the past 17 years truly intuitive, easy to manipulate, and approachable. The learning curve is still perilously steep - for us as an industry to grow any larger, we will have to pay attention to the coming class of young animators that don't have the years of growing up with innovation like we did - they are forced to learn it all, quickly, so that they can get into the food chain - unfortunately it is not that easy.

Pete Docter, director of Monsters Inc., expects that such changes will lead to people with live-action film skills being able to move into animation. Thus lighting experts will be able to design the lighting on future animated films, whereas it currently takes a computer science expert. (74)

In any case, CGI is clearly here to stay and its use, in live-action feature films at least, is likely to increase. It has revolutionised special effects, virtually eliminating the use of traditional matte painting for special effects (75) and drastically reducing the need for stop-motion model work on films such as The Lord of the Rings or the Star Wars series. (76)

As far as animation is concerned however, uncertainties remain. Stop-motion for cartoon feature films has never been healthy and is certainly under threat. Studios are unlikely to invest in it until success seems more guaranteed. Aardman's partnership with Dreamworks is surely the key to this. While Aardman show every sign of being able to continue making successful short films, its problems with The Tortoise and the Hare sound a warning note. A successful second feature for Dreamworks - and one starring Wallace and Gromit has a good chance - could cement the future of stop-motion features.

Those within the industry seem convinced that although 2D cel animation will survive, it will be largely in the form of hybrid 2D/3D cartoons. As well as reducing costs, using CGI for backgrounds allows for a more dynamic camera, matching the increased demand for thrills and excitement. In any case, the training offered to would-be animators these days is heavily biased in favour of CGI, so young artists with traditional 2D cel skills are becoming harder to find. (77)

While Tippet's DID (see the chapter on the Impact of CGI on Stop-Motion, p.11-12) had only a brief life, we can expect exotic haptic devices to become cheaper and more prevalent as computer power increases, just as mice, tablets and scanners already have. (78) Such devices will make the construction and manipulation of CGI models more physical - more like throwing a pot on a wheel than writing a computer program.

As more and more everyday objects become computerized (the average new car now contains at least a dozen tiny computers) it becomes both possible and cheap to make fully computerized, miniature models. We might expect to see a revival of stop-motion as the process of animating such models become simpler and more intuitive. As an example of what I mean, consider Stuart Little. Currently animators have to choose between a virtual, CGI-generated Stuart on the one hand, or a stop-motion model Stuart on the other. In the future one can imagine a real, robotic, model - an intelligent puppet - that exhibited the same range of expression and movement and did so in a semi-autonomous fashion. (79) Where today, reshooting stop-motion scenes is a labour-intensive, completely manual process, such devices would make it possible to design and record such motions and then repeat them, with variations, at will. Stop-motion would effectively be transformed into live-action. (80)

While audiences have eagerly welcomed CGI toys, insects, monsters and aliens, they do not seem ready to embrace believable human CGI characters. If Final Fantasy had had a more compelling story things might have been different, but the losses incurred by Square are likely to discourage another attempt in the near future.

Pixar, with four very successful features to its credit, seems unlikely to fail either technically or artistically. In John Lasseter it has a master animator who understands how to use computers in the service of story telling and the company has others (Pete Docter, Brad Bird) who are almost as capable. (81) Dreamworks SKG's intense rivalry with Disney and PDI's demonstrated ability to make very successful animated features will ensure that Pixar does not have a monopoly of success. It is likely, however, that at least some of the many CGI cartoons currently planned or in production (82) will turn out to be failures, but that has always been true of filmmaking.

Let me end with two quotes that seem to sum up the likely future. First, from John Musker, director of Disney's soon-to-be released 2D/3D hybrid, Treasure Planet:

Some people say it's like sound versus silent films. I hope it's not like that, but some of this is driven by the public. The trick is, the executives' "reading " of the public's desire, so if they feel they can only be successful with CG, then it's a self-fulfilling prophecy. ... I think 3D films are going to go through a phase. I think a lot of studios will get on the bandwagon and there are going to be some bad ones out there. (83)

And finally, from John Lasseter, for me the 'sine qua non' of CGI cartoons:

In about 2 years there will be a series of two to three computer animated films that will not do well at the box office and a hand-drawn film that is great and will rake it in. The headlines will be 'The Resurrection of Hand Drawn Animation. Computer Animation is Dead'. It is in the hands of the artist, and animation is a wonderful thing. I actually think it can do far more than we've ever seen. Both puppet, computer and hand-drawn will co-exist well. Basically what makes money is entertaining films. If it entertains, it fundamentally has the best chance of making money at the box office. (84)

Appendices

The following four appendices contain material that while inappropriate to place within the body of the report, complements and supports its arguments. The first appendix, on Moore's Law, Hardware Technology and Computer Graphics, discusses the hardware and software trends that lie behind the rapid evolution of CGI. The second gives box office and budget figures for most of the films discussed in the report. The third has brief biographical notes on five key figures in the recent history of animation - Tim Burton, John Lasseter of Pixar, Peter Lord and Nick Park of Aardman and Henry Selick. The fourth and final appendix contains a list of forthcoming 3D CGI Films.

Appendix 1 - Moore's Law, Hardware Technology and Computer Graphics

Craving for excitement is the last symptom of a starved imagination.

- Elizabeth Beardsley Butler, 1909 (85)

Moore's Law

The continued increase in the number of transistors that can be fabricated on a single 'chip', a trend known as Moore's Law, been the primary driving force in the astonishing changes in computer performance for more than thirty years. In 1965 Gordon Moore, co-founder of Intel, predicted that the number of gates (transistors) that could be put on a single integrated circuit would double every eighteen months, or put another way, would go up by a factor of 1000 every 15 years. (86) The 'law' has held good for over 35 years. (87)

Intel Processor

Year of Introduction

Transistors

4004

1971

2,250

8008

1972

2,500

8080

1974

5,000

8086

1978

29,000

286

1982

120,000

386

1985

275,000

486 DX

1989

1,180,000

Pentium

1993

3,100,000

Pentium II

1997

7,500,000

Pentium III

1999

24,000,000

Pentium 4

2000

42,000,000

Processor Complexity

As gates get smaller, they operate faster, since the signals have less distance to travel. Since the fabrication cost of an integrated circuit is not directly related to the number of gates (in essence the process is a kind of photographic etching), this means that while hardware prices have remained roughly constant, there have been large increases in performance. If we compare a personal computer of 18 years ago with one of today:

1984

Today

Ratio (now : then)

Price

£1000 or more

£1000 or less

1:2 or less

Processor speed

4MHz

2Gz or more

500:1 or more

Disk

10MB

50GB or more

5000:1 or more

Main Memory

256K

256MB or more

1000:1 or more

Graphics Memory

16K

16MB or more

1000:1 or more

Home Computer Comparison

Moore's Law has been directly responsible for the reduction in the number of components needed to build electronic devices. This manifests itself in the dramatic reduction in the size and weight of products such as mobile phones as well as the very existence of home computers.

There are of course real physical limits:

There are several more 18-month cycles to go before we hit these limits. Most experts, including Moore himself, expect Moore's Law to hold until around 2017.

Computer Graphics ...

The early history of Computer Graphics begins with the work of Ivan Sutherland and Douglas Engelbart in the 1960's. (88) Early hardware was extremely expensive and restricted to large corporations where it was mainly used it for CAD (Computer Aided Design). These were monochrome vector-graphic systems, quite unsuitable for photo-realistic rendering. By the late 1970's and early 1980's, bitmapped raster-graphic systems with colour displays became much cheaper as chip densities increased. Rendering software, originally developed for flight simulators, became more and more photo-realistic as hardware prices came down and better software algorithms were developed. Without the spectacular changes in hardware, today's graphics would still be much like those seen on computer displays in 2001: A Space Odyssey, which were state of the art in 1968.

... Motion Blur and Particle Systems ...

These two frames (actually two frames apart) from A Bug's Life demonstrate both motion blur and the use of particle systems. Flik's ingenious harvesting machine has a circular saw at the front that is blurred because it is rotating rapidly. The hundreds of wood chips, which fly off as he cuts, are also blurred. Whirling as they move, they follow individual trajectories under the influence of gravity and are properly affected by air resistance, in accordance with Newtonian physics. A 3D stop-motion film could not reproduce this scene, while in a conventional 2D cartoon a few simple flowing lines would represent the motion of the wood chips.

... and Computer Games

Moore's Law may have made developments in hardware and computer graphics possible, but there was relatively little pressure for realistic graphics from business until the middle of the 1980's. Engineering design, particularly within the aerospace and automotive industries, was an early but specialised adopter, while advertising/media, now major users, hardly used computers at all until machines like the Quantel Paintbox appeared (early 1980's) and when PageMaker and a laser printer became available on the Macintosh some years later. What did, and still does, force change was the electronic games market, which has three main components:

  1. the special-purpose game machine, starting in the 1970's with Pong and Space Invader machines in pubs, leading to today's games arcades full of specialised machines that offer a single game
  2. home game consoles such as Sega's Dreamcast, Nintendo's GameCube, Sony's PlayStation and Microsoft's X-Box. These machines are essentially high-performance computers with hardware graphics engines to boost their performance and several have DVD drives.
  3. games that run on (high-end) home computers

Games themselves fall into several categories, including:

Of course, some games are hybrids. For example, Microsoft's Age of Empires is an excellent combination of a 'god' simulation with a military strategy game. Similarly the Lucas Arts Indiana Jones games are combinations of adventure games and shoot-em-ups.

1990

1992

1997

2000

All games have become more and more graphical and realistic. For example, where early adventures like Zork were text-based and hence depended, like radio drama, on the player's imagination, the latest are increasingly film-like. The images on the left, taken from Lucas Arts Monkey Island series, are a good example of how the state of the art in real-time, cartoon-like animation has progressed.

As the computational power of game platforms rapidly increases, the sophistication of software for interactive entertainment expands in response. The gulf between high-end graphics, as seen in feature films, and state-of-the-art games, such as Halo for the X-Box, is diminishing. Not surprisingly both Lucas, through Lucas Arts, and Spielberg, through Dreamworks Interactive, have interests in the games market.

For a clear illustration of the close relationship between graphics, games and animation, here are extracts from the Call for Papers for GAME-ON 2002, the 3rd annual European Conference on Simulation and AI in Computer Games, to be held at the University of Westminster in November 2002: (89)

Artificial Intelligence:

Designing (Extensible) AI Engines with Built-in Machine Learning Technologies, Using Adaptive Markov Models,Using Decision Trees, Production Rules and Learning, Using Fuzzy Logic for membership functions and inference procedures, Using Rule Based AI or a Finite State Machine (FSM), Using Fuzzy State Machines (FuSM) or Cascaded FuSMs, Using Artificial Life and layered AI Techniques, Level-of-Detail AI, Using scripting languages to govern NPC Bots, synthetic characters, or believable agents, Controlling simulated characters (Group Behaviour control) using f.ex. flocking algorithms based on extensible scripting systems, Cognitive Modeling: (combining geometric models and inverse kinematics to simplify key-framing. physical models for animating particles. Bio-mechanical modeling, behavioral modeling), Domain knowledge specification and character instruction, Creating AI Networks using supervised learning and genetic algorithms, and pathfinding, Using Databases using the winnowing algorithm, Using Multi-user Data Management

3D Scalability:

MRM (Multi-Resolution Mesh) Technology and the Messiah and Lith Tech Engines; Scalable level of detail-oriented rendering; Methods for scaling animation quality; Scaling animation quality, new animation steps, on interpolated key-frame animation or key-frame morphing; Bump mapping: emboss-dot product and environment mapped bump map (EMBM).

Facial Animation:

Facial animation for Real-Time, Model Behaviour of 3D Modeling; Modelling the bone structure of faces

Skeletal animation and fully scaled rendering:

Physical Simulation, 3D Character Animation and physical controllers; Simulation performance; Rigid body physical animation and rigid body dynamics; Polygon Character Design and level of Detail under Technical Constraints; Particle systems, full polygonal models or sprites; Smooth rendered skins, soft skinning, head animations and full body animation (Skin, extrude and boolean, Design, composition and anatomy); Skeletal, skinning, single skin meshes; Creating Character Animation Assets; Real-Time motion Synthesis, Kinematics and Dynamics, Animating the real-time run cycle; T-Buffers and motion blur; Motion Capture Techniques

3D in Game Animation:

Creating and scaling special effects in Real-Time 3D: environmental weapon effects and general pyrotechnics, software used to produce single frame and animated textures, booth looping and linear, and the pivotal role of alpha channels. Modeling an animation of the geometry needed and the system used to encode additional engine-specific timing and trigger data into the files. The use of the engine particle system and scripting capabilities, Weighted vertices, Streaming SIMD Extension Overview (floating point instruction); Pre-rendered cinematics; Scaling of special effects and texture tricks: particle systems for generating smoke and fire, texture tricks, for volumes, lens flares and onscreen pyrotechnics, Animation Blending

Tools:

Silicon Graphics (MAYA, as a game prototyping environment), 3D Programming for Rage Programmable Shaders (Renderman)

Rendering:

Rendering Equations and architectures; Image Based Rendering (polygon counts (throughput) and overdraw (filtrate); Photorealistic rendering using Open GL and Direct 3D; Multi texture tricks like gloss mapping, dynamic environment mapping, detail texturing and bump mapping Spatial aliasing and Anti-aliasing and accumulation buffers; Setup, Rendering and Transforms; Full floating point setup; Perspective-corrected texture mapping, multiple filtering modes, sophisticated texture blending for special effects and effective looking transparency; Classical local illumination equations and colour theory; Creating Reflections and shadows with stencil buffers and Z-Buffers; Light maps and changing texture coordinates, shadow maps, projected shadow maps; Methods for scaling lighting and shadows, lighting calculations; Equation on a per pixel basis, pixel path and voxel animation; Procedural Texture Methods and Theory and Real-Time; Procedural Texture Implementation; Parametric Surfaces, Deforming surfaces, Curved surfaces and tri-linear flip-flopping Using NURBS (non-uniform rational B-splines) and other parametric surfaces for representing 3D Geometry; Matrix Manipulations; Methods for scaling geometry using parametric curves and surfaces in relation to polygonal models; Progressive meshes and subdivision surfaces

Appendix 2 - Box Office and Budget Figures for Animated Features

 

Box office figures for released films can be misleading. While US (domestic) box office figures are usually reliable, world-wide figures are much harder to find and check. Even where domestic figures are concerned, video and DVD sales and rentals have not usually been included though these clearly affect the financial success of a film. (90) Judging financial success is even more difficult, as studios often try to keep such information secret, using accounting tricks to inflate or reduce announced budgets.

The following table uses figures taken from the www.the-numbers.com and www.boxofficemojo.com websites, cross-checked with other sites in most cases. The budget figures typically cover just production costs. Note that Dreamworks spent $45 million marketing Chicken Run, more than its production budget!

Title

Release Date

Budget

World-wide

Box Office

A Bug's Life

1998

$30 million

$358 million

Antz

1998

$60 million

$152 million

Chicken Run

2000

$45 million

$223 million

Dinosaur

2000

$200 million

$355 million

Final Fantasy: The Spirits Within

2001

$137 million

$32 million

Ice Age

2002

$65 million

$331 million

James and the Giant Peach

1996

$38 million

$29 million (US)

Jimmy Neutron

2001

$25 million

$81 million

Mars Attacks!

1996

$70 million

$38 million (US)

MonkeyBone

2000

> $20 million (91)

$5.4 million

Monsters, Inc.

2001

$115 million

$523 million

The Nightmare before Christmas

1993

$23 million

$50 million (US)

Shrek

2001

$50 million

$482 million

Titan A.E.

2000

$85 million

$30 million (US)

Toy Story

1995

$30 million

$356 million

Toy Story 2

1999

$90 million

$486 million

Appendix 3 - Biographical Notes

Five men, Tim Burton, John Lasseter, Peter Lord, Nick Park and Henry Selick, have played key roles in the developments that lie at the heart of this research report. Given their importance in the recent history of animation, it is worth looking for clues in their personal histories.

Tim Burton

Tim Burton was a student on the first Disney-sponsored character animation program at the California Institute of the Arts (Cal Arts) where his classmates included John Lasseter, Brad Bird and Henry Selick. (92) While still a student at Cal Arts he started working at Disney, first as an animator and then as a character designer. While at Disney, Burton made a six-minute animated short, Vincent, narrated by Vincent Price. The film used the skewed perspectives of German expressionism to portray the dual life of a tortured, but seemingly normal suburban child; it won a number of awards and was released commercially in 1982. His next venture, the 29-minute live-action film Frankenweenie (1984), was deemed such an unsuitable Disney product that it did not receive a proper release until 1992. It did lead to his first directing assignment, Pee Wee's Big Adventure (1985). This was followed by Beetlejuice (1988), and the enormously successful Batman (1989). Burton has since made a series of intensely personal films: Edward Scissorhands (1990), Tim Burton's 'The Nightmare Before Christmas' (1993), Ed Wood (1994) and Mars Attacks! (1996). His most recent film was Planet of the Apes (2001), a loose remake of the 1968 science fiction classic.

John Lasseter

In 1978 John Lasseter left high school to enrol on the same course as Tim Burton and Henry Selick. In 1982 he joined the Disney studio as an animator. Not long afterwards Disney began work on the first film to use computer animation, Tron. Lasseter was able to get an early glimpse of the film's 'light cycle' sequence and says:

It absolutely blew me away! A little door in my mind opened up. I looked at it and said, `This is it! This is the future!' (93)

He persuaded Disney to let him do a thirty-second test that combined hand drawn animation with computer backgrounds. Lasseter recalls:

It was exciting, but at the time, Disney was only interested in computers if it could make what they were doing cheaper and faster. I said, `Look at the advancement in the art form. Look at the beauty of it.' But, they just weren't interested. (94)

Soon afterwards Lasseter left Disney for Lucasfilm, where Edwin Catmull (a key figure in the history of computer graphics, who is now Pixar's chief technology officer) was starting up a computer division. In 1986, Steve Jobs, co-founder of Apple, purchased the computer division of Lucasfilm and incorporated it as an independent company, under the name Pixar, where he now serves as chairman and chief executive officer.

Over the next decade, the Pixar studio, located in Point Richmond, California, led the computer animation industry both technically and aesthetically. Lasseter directed the studio's first short film, Luxo, Jr. in 1986. Two years later, another of the studio's shorts, Tin Toy, also directed by Lasseter, became the first computer-animated film to win an Academy Award. Lasseter's shorts, shown each year at SIGGRAPH, established him and Pixar in a pre-eminent position in the rapidly developing world of CGI. Pixar worked on a large number of advertisements before producing their first full-length cartoon feature, Toy Story directed by Lasseter and released in 1995. Since then the studio has produced three more features and has several more in various stages of development.

Peter Lord (95)

Whilst still at school, and then university, Peter Lord and his partner David Sproxton began contributing short animated films to a BBC television programme for deaf children called Vision On. Heavily influenced by the work of Ray Harryhausen, they eventually focused on clay animation, mainly because nobody else in Britain appeared to be working in the medium (which gave it a saleable uniqueness) and because it offered a flexibility denied to the most sophisticated puppet. In the late 1970s they made a series of successful shorts - first Animated Conversations for BBC and then Conversation Pieces for Channel 4. The success of these began to generate commercial work that, in turn, required the employment of more animators. Among those who joined the studio were Barry Purves and later Nick Park, who was invited to join Aardman in order to complete A Grand Day Out. Following the success of the Wallace & Gromit shorts, Aardman signed a deal with Dreamworks to produce five feature-length animated films. The first of these, Chicken Run, directed by Lord and Park, was released in 2000. Work is currently underway on two more films, The Tortoise and the Hare and a Wallace & Gromit feature, The Great Vegetable Plot.

Nick Park

Nick Park began his career creating animated films while in his early teens. One of his early efforts, Archie's Concrete Nightmare, was broadcast by the BBC in 1975, when he was just 17. While still at the National Film and Television School he began work on A Grand Day Out, a stop-motion clay animation short featuring Wallace, an eccentric inventor with a love of cheese, and Gromit, his faithful yet put-upon dog. When Park joined Aardman in 1985, he simultaneously worked on completing this film as well as on other projects. He worked on the Peter Gabriel award-winning music video Sledgehammer (1986) with the Brothers Quay and Aardman co-founder Peter Lord. Derived from a series of commercials, Creature Comforts (1990), which depicted unhappy zoo residents commenting on and complaining about their climate, diet and accommodations won him an Oscar. One of the other nominations that year was A Grand Day Out which had finally been completed and broadcast on television. This was followed by The Wrong Trousers (1993), which earned Park his second Oscar. Following this success the BBC commissioned him to produce the third in the series, to be aired at Christmas 1995. With just 18 months to produce the film, Park ceded some control and worked with a staff of over 25 camera operators, animators and model-makers. For the first time he functioned more as a director than an animator. The result, A Close Shave, won Park his third Oscar.

Henry Selick

After graduating from Cal Arts in 1977, Selick at Disney worked as an in-betweener and animator trainee on such films as Pete's Dragon and The Small One. He became a full animator on The Fox and the Hound and worked on the two title characters. He also became a close friend of Tim Burton.

In the middle of the 1980s he left Disney and started his own production company. He soon began producing a series of memorable station ID's and 'top-of-the-hour' segments for MTV including the Award-winning Haircut M campaign that featured a stop-motion creature carving the MTV logo in a red Eraserhead hairdo. In 1990, Selick created a six-minute film,Slow Bob in the Lower Dimensions, sponsored by MTV. Combining a live-action central character with stop-motion and cut-out animation, the six-minute film gained first place at the Ottawa Animation Festival and second at the Chicago Animation Festival. It also led to Selick's reunion with Burton and his role as director of the first full-length stop-motion feature film from a major studio, The Nightmare Before Christmas.

Selick's 'reward' for the success of Nightmare was to film the adaptation of James and the Giant Peach by Roald Dahl. Released in 1996, the film did not repeat the success of Nightmare. A projected film for Miramax, Toots and the Upside Down House, to be scripted by Steven Soderbergh, was, like James, to have a stop-motion centre framed by live action pieces. It was shelved due to cost considerations and Selick took up some smaller, personal projects before returning to features with Monkeybone (2000). A mixture of live action and stop-motion animation, Monkeybone also met with very limited commercial success.

Appendix 4 - Forthcoming 3D CGI Features (96)

Project

Type

Studio

Status

Treasure Planet

2D/3D Hybrid

Disney

Awaiting Release

Sinbad

2D/3D Hybrid

Dreamworks

Production

Delgo, A Hero's Journey

All CG

Fathom Studios

Production

Axis

All CG

Chaman & Namco

Production

Living Forest

All CG

Dygra

 

My Peoples

2D/3D Hybrid

Disney

Production

Greanwold

All CG

Greanwold

Hiring/Production

Thru the Moebius Strip

All CG

GDC Technology

Hiring/Production

Jonah and the Whale

All CG

Big Idea

Hiring/Production

Gaina

All CG

Coscient & Chaman

Hiring/Production

Finding Nemo

All CG

Pixar

Pre-production

Water Warriors

All CG

Silver Fox

Hiring/Pre-production

Teenage Mutant Ninja Turtles

All CG

Digital Rim

Pre-production

Valiant

All CG

Ealing Studios

Pre-production

Curious George

All CG

Universal, Imagine & ILM?

Pre-production

Madagascar

All CG

PDI/Dreamworks

Pre-production

Chicken Little

All CG

Disney Burbank, CA

Pre-production

Shark Slayer

All CG

PDI/Dreamworks

Pre-production

Dragons Lair

2D/3D Hybrid

Don Bluth Films

Pre-production

Looney Tunes 3D

3D/Live Action

Warner Bros

Pre-production

LEGO Movie

All CG

?

Development

Over The Hedge

All CG

PDI/Dreamworks?

Development

Gomeo and Juliet

3D/Live Action

Disney

Development

Bromeliad Trilogy

3D/Live Action

PDI/Dreamworks

Development

Unannounced

All CG

Wil Vinton& Kleiser-Walczak

Development

Ant Bully

All CG

Universal & Playtone & ILM?

Development

Route 66

All CG

Pixar

Development

Gulliver's Travels

All CG

Mainframe

Development

Shrek2

All CG

PDI/Dreamworks

Development

Rendevous with Rama

All CG?

Revelations Entertainment

Development

The Incredibles

All CG

Pixar

Development

Polar Express

All CG

Sony

Development

American McGee's Alice

All CG

Dimension

Development

Mindbridge

All CG

Blur

Development

Mighty Mouse

All CG

Nickelodeon & DNA

Development

Automation

All CG

Crest & Columbia

Development

Bloodquest

All CG

Exile Films & Games Workshop

Financing

Foodfight

All CG

Threshold Digital

Financing

Tusker

All CG

PDI/Dreamworks

Dead

Astroboy

All CG

Sony

Development

Vortex

All CG

Foundation Imaging

Development Hell

Unannounced Dean Devlin 3D Project

All CG

Centropolis & Sony

Development Hell

ElfQuest

All CG

Wolfmill Entertainment

Development Hell

Where the Wild Things Are

All CG

Universal, Playtone & ILM?

Development Hell

References

Films

A Bug's Life (Lasseter, 1998)
Antz (Darnell/Johnson, 1998)
Chicken Run (Lord/Park, 2000)
A Close Shave (Park, 1995)
Final Fantasy: The Spirits Within (Sakaguchi/Sakakibira, 2001)
A Grand Day Out (Park, 1990)
Ice Age (Saldanha/Wedge, 2002)
James and the Giant Peach (Selick, 1996)
Jimmy Neutron - Boy Genius (Davis, 2001)
Mars Attacks! (Burton, 1996)
MonkeyBone (Selick, 2000)
Monsters Inc. (Docter, 2001)
Shrek (Adamson/Jenson, 2001)
Stuart Little (Minkoff, 1999)
The Sandman (Berry, 1991)
Tim Burton's The Nightmare Before Christmas (Selick, 1993)
Titan A.E. (Bluth/Goldman, 2000)
Toy Story (Lasseter, 1995)
Toy Story 2 (Lasseter/Brannon, 1999)
Tron (Lisberger, 1982)
The Wrong Trousers (Park, 1993)

Journal Articles and Books

Radio Interviews conducted by Max Schönherr for German National Public Radio:
  These are available in Real Audio format from www.maxschoenherr.de

Web Sites

us.imdb.com
www.aardman.com
www.animationartist.com
www.animationmagazine.net
www.awn.com
www.boxofficemojo.com
www.chud.com
www.digitalanimators.com
www.heweek.co.uk
www.highend3d.com
www.ilm.com
www.matchmove.com
www.maxschoenherr.de
www.pdi.com
www.pixar.com
www.stopmotionanimation.com
www.the-movie-times.com
www.the-numbers.com

Newsgroups

alt.movies.visual-effects
comp.graphics.animation

Footnotes:

  1. Although much of my work involved networking software, in 1986 Addison-Wesley published a book I wrote on the design and programming of Adventure Games. (back)
  2. Though it was brilliantly directed by Henry Selick, Tim Burton's name in the title reflects his role as the major creative force behind the film. He created the story and much of the artwork ten years before while working at Disney. See Appendix 3 for biographical notes on Burton and Selick. (back)
  3. The first feature film to make a showcase of computer generated images was Tron (Lisberger, 1982), although it was not until Terminator 2: Judgement Day (Cameron, 1991) that CGI as we know it today played a major role in a film. (back)
  4. Appendix 3 contains biographical notes on John Lasseter and Pixar. (back)
  5. The seminal paper on the subject is Bower, J. L. and C. M. Christensen (1995), 'Disruptive Technologies: Catching the Wave', Harvard Business Review January-February 1995, 43-53. (back)
  6. There is some discussion of this within the text, but Appendix 1 contains fuller treatment of the technical issues. (back)
  7. See Appendix 2 for box office figures and budgets for most of the films discussed in the report. (back)
  8. Eric Schneider, 'The Art of Ladislaw Starewicz', Kabinet 4, Autumn 1998. (back)
  9. See Richard Taylor, Encyclopedia of Animation Techniques, (London: Quarto, 1996), 10-12. (back)
  10. In the assembly-line method of cartoon production refined by Disney Studios in the early 1930s, and later copied by all its competitors, the position of character designer was of utmost importance. See Gene Walz, 'Charlie Thorson: Character Design In Classic Animation', Animation World 2.6, September 1997 for an illuminating profile of such a specialist. (back)
  11. See Kristin Thompson and David Bordwell, Film History: An Introduction (New York: McGraw-Hill, 1994), 386-388. (back)
  12. Because of commercial breaks, the actual length of such US-produced cartoons was around 22 minutes. (back)
  13. There have been 'portmanteau' features from other studios, e.g. Warner's The Bugs Bunny/Road Runner Movie (1979). (back)
  14. Willis O'Brien (King Kong) and his apprentice, Ray Harryhausen (Jason and the Argonauts) were the two masters of this branch of stop-motion animation. (back)
  15. See Appendix 3 for biographical notes on Peter Lord and Nick Park. (back)
  16. Jaron Lanier, 'Taking Stock', Wired 6:01 (January 1998). (back)
  17. Barry Purves, ' Boldly Throwing Down the Gauntlet', Animation World Magazine 2.11, February 1998. (back)
  18. Tools for CGI modelling include Maya, LightWave and 3D Studio Max, all of which allow popular shaders to be used as 'plug-ins'. (back)
  19. For an illustration of RenderMan in action see Plates II.21-II.37 in Foley and van Dam, Computer Graphics: Principles and Practice 2nd Edition, (USA: Addison-Wesley, 1990). Chapters 14 and 16 describe the algorithms involved in detail. (back)
  20. See Barbara Robertson, 'Toy Story: A Triumph of Animation', Computer Graphics World, August 1995. (back)
  21. This had grown to 1400 by the time Toy Story 2 was released - see Barbara Robertson, 'The Toys are back! (Technology in Toy Story 2)', Computer Graphics World, November 1999. (back)
  22. This, rather than lack of ambition, is why it took Pixar a decade before they could move from shorts to features. They had to wait for Moore's Law to reduce rendering times significantly. See Appendix 1. (back)
  23. The purely digital Yoda in Star Wars Episode II: Attack of the Clones (2002) is a real achievement in this respect. See Barbara Robertson, 'Attack of the Clones', Computer Graphics World, June 2002. (back)
  24. Craig Reynolds, 'Flocks, Herds and Schools, A Distributed Behavioural Model', Computer Graphics 21, July 1987. (back)
  25. The chase scenes in Toy Story for example, or in A Bug's Life. (back)
  26. Bill Hunt, 'Pixar talks A Bug's Life - interview with Bill Kinder and Leo Hourvitz', Digital Bits, April 1999. (back)
  27. For example, the moving vehicles in the road-crossing sequence in Toy Story 2 are severely blurred. For a graphic example, see the frames from A Bug's Life at the end of Appendix 1. (back)
  28. Puppets can rarely be in motion during an exposure unless they are travelling in moving vehicles, as in the train sequence in The Wrong Trousers (Park, 1993). (back)
  29. Gabriel J. Brostow & Irfan Essa, 'Image-Based Motion Blur for Stop Motion Animation', SIGGRAPH 2001. (back)
  30. Stephen Arthur, 'Voila! Masters reveal their methods', North West Animator, September 1997. (back)
  31. See for example Foley and van Dam, Computer Graphics: Principles and Practice 2nd Edition, (USA: Addison-Wesley, 1990), 1031-1034. (back)
  32. See the frames from A Bug's Life in Appendix 1 for an example of the use of particle systems. (back)
  33. Fire and smoke in Nightmare are handled this way. (back)
  34. Barry Purves, 'The Emperor's New Clothes', Animation World 1.1, April 1996. (back)
  35. Barry Purves, 'Boldly Throwing Down the Gauntlet', Animation World 2.11, February 1998. To be fair, he preceded this by saying that 'the wonderful characters in James and the Giant Peach would not have worked as well in CG'. (back)
  36. As a largely manual task, stop-motion costs are directly related to wages. However see the chapter on Conclusions - Likely Futures for some ideas on how stop-motion might evolve. (back)
  37. Stephen Arthur, 'Voila! Masters reveal their methods', North West Animator, September 1997. (back)
  38. Barry Purves, 'Boldly Throwing Down the Gauntlet', op. cit. (back)
  39. See for example Ed Hooks, Brad Bird, Mike Caputo, Acting for Animators, (London: Heinmann, 2001). (back)
  40. Scott Tobias, 'Aardman Animations', The Onion a.v. club 36:23, June 21, 2000. (back)
  41. For more details, see Rod Bennett, 'Jurassic Park and the Death of Stop-Motion Animation', WONDER #9, 1994. (back)
  42. Haptic devices allow the manual exploration of virtual environments and the manipulation of virtual objects. We shall return to this point in the chapter on Conclusions - Likely Futures. (back)
  43. Jody Duncan, 'Flesh for Fantasy', Cinefex, 86: 2001, 43. (back)
  44. Box Office and budget figures for most of the films discussed in this report can be found in Appendix 2 (back)
  45. Details of the making of Mars Attacks! can be found in Wendy Jackson Hall, 'Mars Attacks!', Animation World 1.10, January 1997 and in Barry Purves, 'The Emperor's New Clothes', Animation World 1.1, April 1996. Appendix 3 contains biographical notes on both Burton and Selick. (back)
  46. Wendy Jackson-Hall, 'Peaches N' Dreams: Henry Selick's James and the Giant Peach', Animation World 1.2, May 1996. (back)
  47. See Mike Clark, 'The intricate recipe for making Giant Peach', USA Today, 1st December 1998. (back)
  48. See Appendix 4 for a list of more than 40 planned CGI animated features. (back)
  49. Scott Tobias, 'Aardman Animations', The Onion a.v. club 36:23, June 21, 2000. (back)
  50. Ibid. (back)
  51. As a result of the continual rewriting, Michael Caine, who had recorded the part of the Tortoise in 2000, will no longer be able to play the part due to other commitments. (back)
  52. See Wendy Jackson-Hall, 'An Interview With Aardman's Peter Lord', Animation World 2.2, May 1997. (back)
  53. For technical details of how the film was made, see Barbara Robertson, 'Reality Check (Final Fantasy)', Computer Graphics World, August 2001. (back)
  54. Some have seen the film as a 3D 'Anime flick', more appealing to Japanese tastes than to Westerners, though its Japanese box-office performance was even more dismal – less than $2 million (see Screen International, October 2001). (back)
  55. One popular explanation for Disney's decision to create a CGI unit of its own is that Eisner was concerned that Disney might someday find itself in competition with, rather than partnering, Pixar. (back)
  56. Stephen Thompson, 'Chuck Jones', The Onion a.v. club 33:13, April 9, 1998. (back)
  57. Thus Brad Bird, who worked in conventional animation first at Disney, at Turner on The Simpsons and King of the Hill, then at Warner where he directed The Iron Giant (1999), joined Pixar in 2000 where he is now directing a CGI feature. (back)
  58. Fantasia 2000 is omitted because it was largely shown in IMAX theatres, but it did not even recoup its budget. (back)
  59. This is due at least in part to the intense animosity that Jeffrey Katzenberg (the K in SKG) has towards Michael Eisner, with whom he had a famous dispute when he worked at Disney. Dreamworks has several times released films with similar themes just before Disney. Thus Antz was released just before A Bug's Life, Deep Impact just before Armageddon and The Road to El Dorado before The Emperor's New Groove. Lord Farquaad in Shrek is rumoured to be a deliberate caricature of Eisner. (back)
  60. This is the official figure. There have been reports in the trade press that the actual budget was considerably higher. (back)
  61. There has been the occasional exception, such as the X-rated Fritz the Cat (Bakshi, 1971). (back)
  62. This quote appears in http://members.fortunecity.com/disneygallery/TitanAE.html. I have failed to track down the original source of the interview. (back)
  63. There is nothing in Braudy and Cohen's Film Theory and Criticism and just a five-page article by Michael O'Pray in The Oxford Guide to Film Studies. By far the most illuminating academic paper on the subject is Steve Tillis, 'The Art of Puppetry in the Age of Media Production', The Drama Review 43.3 (1999) 182-195. (back)
  64. Thus Top Cat, a popular Hanna-Barbera TV series in the 1960s, was closely modelled on Sgt. Bilko, the 1950s sitcom starring Phil Silvers, both in terms of characterisation and plot. The connection was strengthened by the fact that Maurice Gosfield, who played Doberman in Sgt. Bilko, voiced Benny, Top Cat’s 'lieutenant'. (back)
  65. In the UK, Oliver Postgate and Peter Firmin's animated series such as Noggin the Nog (cut-out) and The Clangers (stop-motion) were made for next to nothing in a garden shed. (back)
  66. Stephen Thompson, 'Chuck Jones', op. cit. (back)
  67. That said, South Park costs less than $500,000 per episode while The Simpsons episode budget is $3 million. South Park makes extensive use of computers to simulate cut-out animation. (back)
  68. Personal communication (July 2002). (back)
  69. Even young children who believe in Father Christmas see a clear distinction between the Disney figures they see in cartoons and the 'impostors' they meet in the Disney theme parks. (back)
  70. Paul Fischer, 'Pete Docter and John Lasseter talk about Monsters, Inc.', Film Monthly, October 2001 (back)
  71. Jonathan Ross, 'Interview with John Lasseter', Regus London Film Festival interviews, November 19, 2001. (back)
  72. As Appendix 1 points out, increases in computer hardware performance allow and encourage advances in software. (back)
  73. Stephen Schleicher, 'Straight Talk - Industry professionals give their views on the industry', Digital Animators, July 2001. Jimmy Neutron, Boy Genius (Davis, 2001) is an example of a film developed by a small newcomer that had real success (and an Oscar nomination). (back)
  74. Max Schönherr interview with Pete Docter on German Public Radio (March 2002), available at www.maxschoenherr.de (back)
  75. See Jonathan Romney's review of The Invisible Art: The Legends of Movie Matte Painting, (Mark Cotta Vaz and Craig Barron) in Sight and Sound, August 2002. (back)
  76. See Barbara Robertson, 'Attack of the Clones', Computer Graphics World, June 2002. (back)
  77. Thus the UK National Film and TV School now offers a 6 month part-time course - 'The Visual Effects Artist and Animators Program' - which offers training in Discreet products such as Combustion, 3DS Max, Flame and Smoke. (back)
  78. For an interesting collection of haptic devices, see http://haptic.mech.nwu.edu/intro/gallery/index.html. (back)
  79. Sony's robotic dog, Aibo, which currently costs around $1000, is an early, primitive example of what I mean here. (back)
  80. Such ideas may seem fanciful, but as Appendix 1 points out, it is probable that in 15 years time computers will be 1000 times as powerful as they are now. (back)
  81. Disney recently offered Lasseter the job of running their animation studio. Steve Jobs countered by giving Lasseter a 10-year contract at Pixar. (back)
  82. See Appendix 4 for a list of almost 40. (back)
  83. See Jack Ruby, 'Interview with John Musker at the August 2002 San Diego International Comic Con Festival', CHUD August 2002. (back)
  84. Jonathan Ross, 'Interview with John Lasseter', op. cit. (back)
  85. Elizabeth Butler, Women and the Trades: Pittsburgh, 1907-08, New York 1909, quoted in Film Before Griffith, J. Fell, (US: Berkeley 1983) 204. (back)
  86. Gordon Moore, 'Cramming More Components Onto Integrated Circuits', Electronics, April 19, 1965. (back)
  87. It is worth pointing out that Moore's Law is not a law at all, but a self-fulfilling prophecy - the semiconductor industry is forced to keep up with it because if it didn't it would appear to be failing. (back)
  88. The standard work on Computer Graphics is still Foley and van Dam, Computer Graphics: Principles and Practice 2nd Edition, (USA: Addison-Wesley, 1990). (back)
  89. The Conference website is at http://biomath.rug.ac.be/~scs/conf/gameon2002/index.html. (back)
  90. As an example of how much revenue can be generated over and above the box-office gross, DreamWorks SKG announced in January 2002 that Shrek had sold over $420 million worth of home videos and DVDs in just over two months on retail shelves. By the middle of 2002 it had also generated more than $51 million in video and DVD rentals. (back)
  91. See Mike Clark, 'The intricate recipe for making Giant Peach', USA Today, 1st December 1998. (back)
  92. Among other students who attended this course, and were taught by Jules Engel, are Eric Darnell (Antz) and Ellen Woodbury (The Lion King). (back)
  93. See Mike Lyons, 'Toon Story: John Lasseter's Animated Life', Animation World3.8, November 1998. (back)
  94. Ibid. (back)
  95. Much biographical material on Lord, Sproxton, Park and other Aardman personnel can be found in Peter Lord, Brian Sibley and Nick Park, Creating Animation - The Aardman Book of Filmmaking, (London: Thames & Hudson, 1999). (back)
  96. This table comes from a list maintained by Gregg Lukomski - see http://www.matchmove.com/projects.html (back)

Back to Film Studies Page