Major Project Report 2003
BA[Hons] Computer Visualisation and Animation
A Little Reverie
By Tessa Mapp
!
For correct formatting please view this html version
This file is
best viewed with Microsoft Internet Explorer
!
CD-ROM
Project report (html format)
Boot shader code (.sl)
Three representative images (JPEG format)
‘A Little Reverie’ (Cinepak format)
Compositing builds (Cinepak format)
Animatic (Cinepak format)
Block test (Cinepak format)
Primary research footage (DivX format)
DivX codec for Microsoft Windows
Abstract
This report describes the production of an animation, which combines realistic elements with non-realistic elements in a montage style, in order to stress an alternative meaning through the visuals.
Aims
-To explore the possibilities of representing nostalgia and memories in a visually creative way.
-To develop a simple narrative incorporating these themes and produce this in the form of a short animation.
Objectives
-To gain a thorough understanding of the themes through research and then use this knowledge to add visual interest to the animation.
-To challenge the current trends in Computer Animation
-Develop a relevant technical understanding in order to produce the
best quality work
-Produce a distinctive look for the rendering style
-Bring these elements together in the production of the proposed story
1 Introduction
2 Film
2.1 Development of story and themes
2.2 Design
3 Ideas into Practice
3.1Timeplan
3.3 Modelling for rendering
3.41 Animation
3.42 Animation Set-up
3.5 Rendering Set-up
3.6 Shader writing
3.7 NPR
3.8 Filming
3.9 Matte Painting
3.91 Miscellaneous
4 Compositing
5 Conclusion
When the camera was introduced in the mid 1800’s it was thought by many that the painting would become obsolete. They feared that the painter would be replaced by photography. Instead of oppressing the painter, it liberated them. Painting became a more expressive art, not trying to recreate nature as a photograph might, but interpret it. As artists we cannot help but be influenced by our culture and manipulated by contemporary events such as inevitable technological advances. Erwin Rosenthal reinforces this idea in ‘The changing concept of reality in art’ saying
The transmutation of artistic form depends on the
individual decisions and cultural developments. [Rosenthal 62] [1]
Similarly when the American radio serial “The Lone Range” (ended in 1951) made the transition to television, nothing was shown on screen without simultaneously being described verbally, continuing the conventions of radio drama. The medium of television had not yet expanded creating its own laws of structure instead borrowing laws that it knew already worked for another medium, radio.
“It has been suggested that the initial content of
every new medium is an old medium *. It
takes time for the inherent potential of any new medium of expression to be understood
and developed.” [Kelly 96] [* Paraphrasing McLuhan 64][2]
“The selective ordering of the Visual is the artists
response to the world’s flux, but it is not atrophied, like the freeze frame of
the snapshot. The retrieval, making a
painting is transmitted through the history of the means.”[Woodiwiss] [3]
The medium of Computer animation has been used in Film and Television for years, as special effects in films, but the medium itself has had little time to develop. In early computer animation, the aim was to create a reality within the computer. Early on, aesthetic rules were set up, based on the need for the medium. Films such as the Abyss (Cameron/Industrial Light and Magic, 1989) and Terminator2 (Cameron/ILM, 1991) pushed the medium towards a photorealistic look where the computer generated images had to blend seamlessly into the reality of the film. In animated short films (Luxo Jr, Pixar 1986), music videos, commercials and more recently animated feature films there has been a development of style to add interest but also as a natural reaction to the wealth of photorealistic work being produced in the post production industry. John Lansdown in ‘Computers & Art’ describes this change in attitude:
Among many artists, however there has been something
of a revolt against photorealism- which they see either as misguided or as far
too restricting to their creativity.
The revolt takes two forms: one, by those who react to existing
developments and wish to investigate computer graphic means of depicting scenes
in non realistic ways; and two by those who ignore depiction of visual reality
altogether and are concerned mainly with abstractions and interactions
[Lansdown 97] [4]
Consequently we are entering a new era for the medium, where the individual style shapes the animation. It was this interest that led to the idea for the animation produced as part of this major project, summed up in a quote from Erwin Rosenthal in ‘The changing concept of reality in art’. Talking about reality:
For it is not anything stable; it is oscillating,
and it comprises not only the visible world but also includes feeling and
emotions. [Rosenthal 62] [5]
2 Film – Photorealism vs.
non-photorealism
The Medium as a Structural agent is often overlooked as Herbert Zettl summarises in 'Sight Sound and Motion’ [6]. He equates a critic who ignores the influence of the medium, to a person saying, “How do-you-do?” but then disregarding whether the person is smiling or frowning. In his example it is clear that we would have a different understanding of the phrase depending on the facial expression of the medium. Computer Visualisation and animation is a new art form. As a medium, it too must mould the communication process.
As we have already discovered we can roughly divide computer animation into two stereotypical groups whose usage has become popular over recent years, photorealism and non- photorealism. When photorealism is adhered to its message as such is ‘believe I exist’. The expression non-photorealism is loosely applied to rendering where the aim is not to reproduce nature but to add some sense of visual style. Its message could also be 'believe I exist' but as an emotional response and not in the literal sense. Craig Reynolds describes the phenomena on his website:
Photorealistic rendering attempts to make artificial
images of simulated 3d environments that look “just like the real world” So non-photorealistic rendering (NPR) is any
technique that produced images of simulated 3d world in a style other than
realism. [Reynolds 02] [7]
Both have their appropriate applications, the first in creating illusions in films, and the second in cartoons or short films. They both have their problematic traits too. Achieving photorealism can be technically challenging and workload heavy especially if you feel that you are trying to create something that already exists. Also whether something is photorealistic depends upon many different skills, modelling, texturing, lighting etc, to produce the effect in films usually a pipeline is used so that people with those skills do that particular job in hand. It is unrealistic to expect employees to be ‘jack of all trades’. Non-photorealism is often challenging to achieve with the software available so usually users resort to writing their own programs. This predicament is approached in a recent paper [HALPER, SCHLECHTWEG, STROTHOTTE, 02] [8]. A modular pipeline, which makes use of elementary operations and adds an artist friendly interface opens the possibility for users with no programming skills to produce their own unique NPR rendering system. By allowing designers to play with ideas the tool should not constrain or influence the natural artistic process. They conclude that the resultant images may not quite be to the users liking and can produce unexpected results. This is evidence that we are not at the stage where the user can have full creative control. Much of the creativity still lies with the programmer. It is a shame that this is so because non-photorealism can have a lot to offer to the digital artist. Instead of bombarding the viewer with distracting fine details, economy of line makes it easier to direct the attention of the viewer. This offers an alternative to the director who would normally have to use other devices to gain the same effect. NPR animations are interesting to watch because it allows the audience to share in the interpretive process.
It is the aim of this project to create a new style by combining the valuable qualities of both photorealism and non-photorealism. It is hoped that this montage of styles will change the meaning of the animation.
[In photorealism] The
quality of images created using a computer is judged by how closely they
resemble a photograph…. In NPR images are instead judged by how effectively
they communicate. When using images to
communicate the essence of a scene, simulating reality is not as important as
creating the illusion of reality. [Gooch 01] [9]
Figure 1. A selection of
images taken from 3D World Magazine (2002-2003). Although all are aiming for realism, it is clear that some of
them still look computer generated. The
images that seem to work best are those that have been used in films or
adverts, such as Gollum from Lord of the Rings (bottom right), the aero
mouse (bottom left) and the sprite (top left).
Photorealism, like pornography, leaves nothing to
the imagination. [Curtis 01] [10]
Rejecting naturalistic forms means a resolution to
change the object subjectively in order to stress its deeper meaning.
[Rosenthal 62] [11]
FilmConcept-
The aim for the project was to present a simple narrative with more complex meaning achieved through the visuals. It was also the aim to give it personal meaning but encapsulating universal themes. These central themes include childhood and music, themes which by themselves evoke feelings of sentiment and nostalgia. A key investigative part of the animation is to communicate these themes through the medium of computer animation applying qualities specific to the medium to express emotions indirectly.
The short piece of music “A little reverie” by Adam Carse gave the inspiration for the initial ideas. As a child I played this piece in a music festival. My cello teacher would tell me a story to help me remember the dynamics of the piece so that I understood how it should be played. The story took me through the emotions felt by the character; I remember that a river with a bridge laid the scene. This has nostalgically stuck with me ever since and led to the development of the child on a bridge. The story also evolved from the idea of a daydream (i.e. the title of the piece) with the music leading into a map and introducing the river.
With the themes of childhood and nostalgia in mind, the process of communication drove the development of ‘A little reverie’. Memories play a significant part in the concept of nostalgia. When we look back on our childhood we seem to remember some people, places, objects, times, better than others. Our memory can play tricks on us and it can be warped by time and emotions. Herbert Zettl describes this selective seeing in ‘Sight, Sound and Motion’
We choose to look at things we like to see and are especially interested in, and ignore those that mean little to us.[Zettl 99] [12]
If things that we are interested in are particularly memorable, then they are most likely to be personal. In this project it was vital to find something applicable to most childhoods, an object that we associate with growing up. The humble Wellington boot, epitomized in E.H. Shepard’s drawings of Christopher Robin in the series of ‘Winnie the pooh’ books, is ideal because of its history. The books themselves are adored by millions of children worldwide, and are known for representing a quintessential age.
Figure 2. A collection of
images showing the history of the Wellington boot.
Synopsis- The exposition scans the music horizontally which moves as if eyes were reading it, turning into a river on a map and establishing the scene of a bridge over a river. On the bridge stands a child. The scene is set in an illustrated world, the only exception to this rule are the child’s red Wellington boots, which are in a realistic style. The game of pooh sticks is established as the child drops a stick on one side of the bridge, crossing the bridge only to see the stick appear on the other side. As the music changes, the twig is shown on an elaborate journey floating past an assortment of discarded items that conjure a sense of history and loss. We then see the child swinging its legs off the edge of the bridge when the boot falls off, into the river. The animation comes to a close as we watch the boot gently float up the river.
Preproduction
Character and set design took place in the very early stages and mostly consisted of simple rough sketches, almost incomplete, conveying an impression rather than outlining an exact description. The final rendering was undecided as the quality and style of the work was unknown. This was to be decided in the shader development and experimentation stage. A few factors influence the character design as follows:
-The innocence of the book illustrations of E.H. Shepard
-For most of the animation only the child’s feet and lower body are visible because the child’s boots are the main character.
-Caricatured to have big boots in relation to the rest of the body in order to emphasise the emotional attachment to them. Betty Edwards describes this in the Drawing on the right side of the brain, illustrated in the images below.
Figure 3. Designs and ideas
for the animation A Little Reverie
Figure 4. In
Drawing on the right side of the brain, Betty Edwards analysis of an
image on the left, drawn by a four year old, describes how the arm that holds
the umbrella is huge in relation to the other arm because it is the important
point in the drawing. Also how the
older sister who has been drawn larger than everyone else and with additional
scary teeth overwhelms a family portrait by a five year old (right). The child
expressed how he felt about his sister through the size of the drawing. pg
66-67
For this project, storyboards were not felt to be relevant as the animation was always designed to be timed to the music. Without the music the story does not visualise as it was intended. As an alternative, an animatic was produced with the storyboard images as a temporal sketch, which was felt to be a much more expressive and illustrative means.
The blocktest eventually replaced the animatic taking the form of low-resolution model playblasts, from Maya, edited together. With the whole animation existing early on in production, work could be replaced as it was updated, obtaining a good idea of how much time was needed for each shot and aided the construction of a schedule for the course of the 8 months allotted production time. It also motivated the production as the blocktest was at one point updated on a weekly basis and it could clearly be seen where the strengths and weaknesses of the project lay. It became compulsory to think about the layout of scene files, setting up camera positions, reasoning the framing, pace, and shot transitions. Keeping the working manner loose gave freedom allowing spontaneity and also enabling the animation to develop and mature.
Figure 5. Storyboard taken
from the animatic of A Little Reverie.
The animatic and the blocktest can be found on the accompanying CD
3 Design- Medium, montage and memories.
Research and influences
The medium of computer animation can be rendered so that it can take any shape on the screen but we are confined to the screen space. Films are usually shown with black bars above and below the animation in order to keep the wide screen (1.85:1) aspect ratio. In photography, the photographer can choose either a portrait or a landscape layout, making use of all the available space depending on the subject matter, but we cannot turn the television or computer screen 90 degrees. In this digital age, how necessary is it to stick to the dimensions of the screen? If the idea for this animation is that we challenge the use of the medium then changing the frame continues this theme. It was felt that a square would frame the image better than a rectangle and so was chosen to give the desired effect. The square frame also references photographs taken on a Polaroid camera which became a novelty of many a childhood, and could be seen as a motif for memories captured with a photograph.
Figure 6. Artificial masking is used to change the
aspect ration in D.W. Griffith’s Intolerace. The effect is to intensify
the falling of the soldier. We have
become so used to watching wide-screen films on television that we hardly even
notice the black bars framing the film. Images taken from Sight, sound and
motion.
Photorealism is a vague and often misinterpreted word. To some it means realism in its most detailed state. To others it means that an image looks like a photograph, i.e. may have depth of field etc. It is a subjective word. Without photorealism there would not be non-photorealism. In achieving a photorealistic look it is important that there is a grounds to do so, for example, why create a computer generated boot if it is easier to film a boot and composite it in? What would it add to the animation? If the boot were to react within the computer generated background, that would be a valid justification. Photorealistic implies a type of perfection and idealisation.
Figure 7. A selection of
work by Cassidy Curtis a pioneer of NPR, researching into many areas of the
field. Here we can see the stark
contrast between edge detection techniques shown top left in a frame from the
animation “The New Chair” and right, in an example of Curtis et al.’s
watercolorization technique[Curts et al 1997].
The bottom left image shows how expandable these techniques are,
displaying a few styles that can be achieved by Curtis’ loose and sketchy
filter.Curtis[Green et al. 99]
To the present day research into NPR has concentrated on edge detection based techniques such as [Green et al. 99] [Gooch et al. 98][Raskar, Cohen 99] or on more painterly effects such as [Strassmann 86][Curtis et al.97][Takagi et al. 99]. This project was never meant to compete with a system set up to create NPR images but they have influenced the direction of the project. The images assimilate a style developed as a mass art i.e. Japanese printing and not something that is individual as a watercolour or oil painting may be. Since animation is a mass produced art, it seems to fit the style of the piece.
Figure 8. Japanese Woodblock
prints by Hiroshige c.1850. In
Japan incompleteness was valued, and mood is emphasised over realistic
depiction. Compositions have
asymmetrical balance, depth is implied through line and not shading.
Indeed sketching and memory are also related, since sketching relies on memory in order to be able to visualise the image.
Gombrich in Art and Illusion states that
We can order the loosest association of marks in
comparison with the memorised schemata and ‘see’ an image [Gombrich 60] [13]
The artist Giacometti also speaking on memory
Memory is short, very short. When you look at reality, it’s so much more
complex, and when you try to do the same thing again from memory, you realise
how little you remember. [Giacometti] [14]
Woodiwiss describes our how our memory works by creating a rough image of our surrounding and then building a complete image by piecing together previous experiences to fill in any unknown gaps. This in itself could be seen as referring to the idea of montage.
In our memory we are able to
retain an accumulation of successive visual cues that build a coherent
image. If we scan a panoramic view, we
trace the whole, each new perception adding to those previously experienced
until we understand the complete image despite that image not being available
to us in total. [Woodiwiss] [15]
Figure 9. A
definition from The Oxford Dictionary of Art.
Montage artists take a collection of images and reform them to create a picture with new meaning. For example, Raoul Hausmann’s photo montage The art critic is unmistakably an anti-establishment work of art, shown through the composition of the individual pieces of photographs. Evidence for this can be seen in the use of a pencil positioned to look like a sword, his weapon, and the money he earns is cut, and place as to be stabbing him. Holes have been punched in his face and sketches placed over his eyes and mouth, defacing him, perhaps suggesting the vicious nature of the art critic, and possibly reflecting a personal vendetta. In ‘A Little Reverie’, the statement is perhaps not so obvious at first due to the personal nature of the montages.
Figure 10. A comparison of the montage technique,
used by Richard Hamilton in ‘Just what is it that makes today’s homes so
different, so appealing?’(far left), Raoul Hausmann’s ‘‘The art critic’
(centre), and in the animation A Little Reveri’. Composition, structure and space need to be
carefully considered in order to produce a statement.
Our conscious memory represses most of our everyday encounters, the events that we remember best have some importance to our lives, and usually there is an emotional tie. Reminders of the past come in the form of photographs, memorable objects, people, places, and can bring a rush of memories back. Sometimes we associate an emotional or traumatic event with an object, we then say that the object has sentimental value. The child in the animation looses a boot in the river; perhaps this will remind the audience of all the objects in the river earlier in the animation, maybe they were lost objects too. Maybe the animation is the child (now grown up) remembering loosing the boot. We often romanticise over the past and with this comes feelings of nostalgia.
The
memory of familiar objects can influence what we perceive. [Gombrich] [16]
In the Jean-Piere Jeunet film ‘Amelie’ (2001) nostalgia is portrayed through various filmic methods. When looking back on the childhood of Amelie in the first sequence in the film, it is almost unnoticeable that it is speeded up as if it is being played on a projector. The entire film is cast with the complementary colours green and red giving the film a warming atmosphere, full of sensation:
Colour has more energy when set off against a
background of contrasting brightness than against a background of similar
brightness. [Zettl 99] [17]
The attention of the audience is directed with the colour to great effect, since the green recedes while the vivid red is prominent and eye catching. Red is also a memorable, and easily imaginable colour, for example it is commonplace to hear ‘when I get mad, I see red!’. This expression also demonstrates how we can relate emotions to colour, other examples include, ‘green with envy’ or ‘I feel blue’. Colours can be exaggerated by memory, and more significantly the emotions of the memory, for example a happy memory may seem more bright, or saturated and an unhappy memory seem more dull.
Figure 11. Still images
taken from the film Amelie (2001).
Boots were chosen for the character because most people have at some
point in their childhood worn or owned a pair of Wellington boots. Here the red boot is shown and she too is
shown on a bridge. The red/green
colouring of Amelie is prominent in these images.
Colour in the piece ‘A Little Reverie’ is harmonious outweighed by the background, which is a yellow-brown paper effect. The addition of the bright red boots introduces a strength and energy associated with childlike liveliness. The bold colour can be seen as emphasising the importance of the boot to the child.
Everyone knows that yellow,
orange and red suggest ideas of joy and plenty
[Delacroix 1901] [18]
In having red Wellington boots
the colour does not discriminate against the sex as perhaps a pink or blue boot
might. By not defining the sex means
that more people can relate to the child in the animation.
Critical analysis
The piece relies on the universal understanding of rhythm. Non-diegetic sound is used in an expressive manner and consists of a musical score, pacing and structuring the animation. Shot transitions and any primary motion is induced by the music coalescing in a unified flow. Connections are made throughout between the animation and music; imagery used with the water is cyclic like the verses of the music, the stave of the manuscript resembles the patterns created by the water, and the musical notes flow into the river, reoccurring in later in shot 10. The musical notes also resemble the shape of the boot. The iterative nature of the animation stresses the relevance of music to the piece.
The plot is narrated through means of a linear objective construction that forms a strong sense of cause and effect. For example the child drops the stick on one side of the bridge, although we are not shown explicitly the journey of the stick as it travels under the bridge it strongly evoked through the linear relation of shots and from past experiences.
As long as the screen
remains flat we should not forget the forms and styles of 2D work [Mackinnon
02] [19]
A little reverie often uses techniques that have their roots in traditional animation. An example of this can be seen in the opening sequence as the music script transmogrifies into the lazy river. Classic Walt Disney fairy tales begin the films with a narrated book introducing and setting the scene for the awaited story e.g. Sword in the Stone (Reitherman, 1963), Sleeping Beauty (1959). The real element acts as the transport between the world familiar to the viewer and the screen space which begins the animation. The morphing technique can be seen in Dumbo (Sharsteen, 1937) when the protagonist hallucinates as bubbles blown from his trunk transform into elephants, also throwing the animation into a completely different narrative.
Figure 12. The map pays
homage to the illustrations of E.H Shepard, children’s books whose influence on
the visual style of the piece has been profound. Illustration is an art which provides explanation.
.
Tried to capture the inquisitive nature of a child. In this sequence the emphasis is placed on stylised illustration using montage to represent the nature of a child’s memory. We watch as the twig is caught in the current of the river, meandering through obstacles that clearly do not belong there intended to be the child’s imagination and intrigue of how the twig got from one side of the bridge to the other. This is echoed in the theme of the entire animation where a non-photorealistic setting is used to portray the simplicity of the surroundings suggesting it is not as memorable as the realistic boot. As if the child has remembered the illustrated book therefore filling in the gaps with what they know to be true. Together they portray a visual representation of memories. There are parallels with the film The Wizard of Oz (Fleming, Thorpe, 1939) where a visual devise is used to recognise the real world from the imaginary, portrayed in sepia tone and full Technicolor. The same device is used in the 1998 film Pleasantville (Gary Ross) where the idealised television show from the 1940’s is portrayed in black and white and the real world from the 1990’s in colour. As the story develops the 1940’s starts to get corrupted by the sins of the 1990’s and colour begins to be incorporated into the black and white world. Here the mixing of the different mediums is used successfully to emphasis the divide in ideological principals.
Figure 13.
Images taken from The Wizard of Oz (1939) and Pleasantville
(1998). Both use the principal of using
different mediums to imply different meanings.
Mobile framing is used to scan the space, horizontally suggesting calmness as the viewer is led into the animation. A similar vertical track is used in a much more powerful fashion when we are following the path of the twig which is led by the melody of the music. The motion of both the twig and the camera in the same direction heightens the power.
Editing techniques used consist of dissolves and fades to black. A dissolve is where two shots are mixed, the first gradually disappears and for a moment two images blend in superimposition. Dissolves are used throughout the piece to slow the motion down, inducing the calming atmosphere of a daydream. Also, dissolves are used to form a strong bond between the two images. An example of this type of montage (in the film sense) is where child walks away; we then dissolve to a shot of a branch swaying in the wind and a twig falling from the branch. This tells the viewer that they are linked generating the image that the child has collected the fallen twig to use in the game. Fading to black adds closure to the shot. Normally this type of dissolve denotes passing of time but the next shot should be in the same time frame. It could be speculated that the fading to black suggests the passage of the twig into the darkness of under the bridge. Perhaps it is not the first time the child has crossed the bridge, in which case the fade to black could still symbolise the passing of time occurring due to repetition of events.
The bridge plays an
important role in setting up the narrative
Figure 14. A comparison
between a water pass featured in the film A little Reverie and a print
taken from a modern Japanese design book
The water takes its influence from Japanese designs and echoing the Japanese woodblock print style adopted for the visualisation of the piece. The direction of the water tries not to detract from the narrative or lead the eye away from the action. In shot 10 the water is flowing down, but in shot 12 it is flowing up. Musically, this shot 12 slows and the pitch becomes deeper, complementing the upward positive motion, likewise shot 10 speeds up and the pitch becomes higher. By both using the same rule of opposites the shots seem to work together even though it goes against our cognition. It could be speculated that the motion of the boot upwards could relate to the feeling of an ‘ending’ like the phrase ‘going up to heaven’. With the filmed and not computer generated boot in the last shot the audience is left to think about why this is so. Perhaps it is the memory of the boot in the final shot, maybe the other objects in the river which were also filmed, are only memories.
Rivers are sometimes seen as a metaphor for life and indeed the animation presents us with an individualistic look at life. If art is to shows us a new perspective on life then this piece should be seen as an artwork because it strives to do so through the stylistic themes. The use of Non-Photorealistic rendering and montage encourages the viewer join in with the interpretive process. The reading of the animation will vary for everyone in the audience because our past perceptions will guide our own interpretations.
3.1Timeplan
Figure 15. A rough schedule
was laid out early on in pre-production (left). As the deadline approached, a more detailed day-by-day plan was
drawn up (right).
Pre-production was given roughly one quarter of the time in order to allow animatic and blocktest to be developed. Animation was seen to be the weakness in the project so research into this came early so that animation could be started in January. Roughly half the time was given to research into the visual style. The remaining time was left for an intense rendering and compositing session and as contingency. Mainly the deadlines were stuck to. Time was lost around other project hand ins but made up for in the large amount of time left for rendering.
The animation was ordered into shots for ease of naming convention. Shots 1 to 12.
The sound was worked closely with the story development. In all it was recorded 3 times in total, once without the piano, for the animatic, and twice for the blocktest, the first time the sound was found to be too fast/slow so was re-recorded and the blocktest was re-timed to fit with the music. This was then used as the final sound. Since the feel for the piece is to accentuate the sentiment to complement the animation it was felt that leaving it raw without any digital enhancement gave a feeling of immediacy associated with older recordings from the past.
3.3 Modelling for rendering
Figure 16. Four views of the
child /character, notice the oversized feet and head compared with the rest of
the body. Also observe the low polygon
count compared with the boots. This is
due to the different rendering pipelines, they do not need to be of the same
level of detail.
Two different styles in rendering require two different styles in modelling. In order to produce interesting NPR the models were left with low numbers of polygons. This gave harsh angular edges throughout which acted as a good base to edge detect and produce the NPR images. Any great detail in the models would be lost in the NPR process. With this in mind it was felt that splitting the modelling time as to give more attention to the boot and less to the scene and child, would be most efficient. In contrast to the rough edges of the surrounding, the boot was modelled to a high- definition state in polygons and then converted to smooth Pixar subdivisions. These are Renderman’s own modelling tools, and are known to give the best results when rendering with Photo-Realistic Renderman (Prman) renderer. Very little artistic interpretation was needed when modelling the boot that was built through direct observation of a real boot. Key features of the model include a fine seem which runs from front to back suggesting that it is a manufactured boot, a similar idea is used in Disney / Pixar’s Toy Story (1995) in the visualisation of the toy soldiers. Also the detailing around the heel of the boot and on the sole, all are key memorable features of the Wellington boot and so would be necessary to portray as realistic a boot as possible.
Figure 17. The boots model
is of a much higher polygon density.
When smoothed with Pixar Subdivisions the model is ready to render.
Modelling in 3 dimensions gives the ability to change lightings and shading of the object but in this animation lighting is irrelevant to the NIZID* shader used in the NPR pipeline. Where there is a need for consistency the scene model is built, for example the bridge and the riverbanks aid shot set-up. Taking the time to model scenery that is un-animated would be inefficient with so few shots. Instead any non-essential scenery was matte-painted.
*(This term is explained later in 3.7 NPR)
Children’s
characteristic + boot characteristics
It was decided to specify a set of animation objectives. These are:
-To capture the essence of childhood, (footage was taken to study).
-To give weight to the character, as it goes up on tiptoes.
-Some anticipation,
rolls back on feet and then up to tiptoes.
-Flexible rubber, squash and stretch in the boot.
-Poses to addressed to the camera.
-Realistic timing and spacing.
-Establish Secondary motion of boot after primary animation has been developed.
Figure 18.
Mini DV camera reference footage can be found on the accompanying CD-ROM
3.42 Animation set-up
Figure 19. Sketches of Lucas
feet enabled understanding of animation in order to set-up the character most
effectively.
The characteristics of the child rely entirely on the animation of the feet. It is therefore vital that the character is set up in such a way not to complicate its actions, but to ease them. From studying the footage of Lucas’ feet, it could be seen that a wide variety of movements most very specific to the movement of the boot would be needed to convey the childlike actions. The boots bend and roll as the child looses balance easily. Children are playful with their feet. Setting the character up at too high a high level (i.e. the commands ‘do walk cycle’, or ‘child loose balance’), would take a long time and would limit the animator’s decisions, but setting the character up at too a low level ie rotate Y about heel joint, would counter an intuitive workflow.
Figure 20.
A balance between the two would be needed to fulfil the criteria. After a few skeleton experiments I decided to use a reverse foot method of controlling the feet. A normal skeleton i.e. knee->ankle->ball->toe can rotate around the ball, toe and ankle but not the heel. Using a reversed shape foot skeleton as a controlling node creates an imaginary heel joint and still gives the same control to the other joints.
Figure 21.
This is achieved by first point constraining the Inverse kinematics (IK) control to the reverse foot ankle control. Then the leg can be controlled by moving the reverse foot. By constraining orientation of the ball joint to the toe control joint and the ankle joint to the ball control joint, the foot can be controlled fully thought the reverse foot bones. For added functionality, NULL groups were created to act as the tilting pivots of the boot. The pivot of the node was moved into a position where the boot would around and not the bone. This tilting on the sides of the boot is a typical action performed by children when playing with their feet. On top of the tilting functionality a final group was created acting as the foot control node. Using the connection editor the required rotations were linked to attributes with more appropriate and descriptive names such as, Toe_rotate, Heel_back_forward, Ball_rotate, etc. To encapsulate the animation controls, unnecessary channels such as scale and rotate were removed and locked so that they could not be tampered with. Limits were also set on rotations e.g. it would not be possible to rotate the ball of the foot more that 45 degrees or rotate the toe sideways more that 180 degrees. This set-up was really useful when it came to animating as there was no decision of which nodes to pick for certain controls. If the right foot was being controlled, the only node to be animated would be the right_foot_root node.
Figure 22.
Figure 23. The boot model
was duplicated and modified according to the sketches produced from the footage
of Lucas (figure 22). Key areas were
identified as being under stress when in certain position. Driven keys could have been used to link the
bones animation to the blendshape animation, but it was felt that this would
take away a lot of creative control from a such a large part of the character
animation.
Blendshapes were created and then attributes added to the foot control node ‘left_foot_root’ and ‘right_foot_root’. By setting driven keys using these attributes, the entire foot control could be accessed through the control nodes.
A
simple skeleton was used for the rest of the body and was bound using Maya’s
Rigid Binding rather than smooth binding because it was found that more control
over specifics could be gained with the weight painting tool.
When painting the bone cluster weights the aim was
to gain a balance between the influence of the leg and the ankle, i.e. the leg
should move around in the boot, but not so much that it could intersect the
boot. This was a frustrating and
laborious task since every shot needed a different weighting. By only weighting boots where it was obvious
that it was needed meant that not too much time was spent tweaking the
weighting, which was cutting into valuable rendering time.
The character had to be set-up differently for shot 11 where the legs dangle off the side of the bridge. The model was modified so as to not have IK in the legs. Animation was then done through rotation of the knee joint, but the foot bones could still be controlled through the reverse foot.
3.5 Rendering Set-up
Figure 24.
Rendering layers is essential for compositing. The Photo-Realistic Renderman (PRman) renderer was chosen in order to gain the most control. From the start of the project it was important that each relevant component be assigned a layer, so that rendering could be started early. Adapters were used so that a user could render the desired layer through the context input, i.e. scene renders the scene etc. The adapter only need be set-up once in the main SLIM palette. The SLIM palette can then be exported for global use. When imported into every scene the correct adapter only need be attached to the object that it is relevant to, and the layers system still stands. Primarily each layer attached the correct shader to the object to be rendered and then used a matte to cut out unwanted areas. This worked fine for the realistic objects like the boot and the sole but when the NIZID channels were edge detected, the resultant lines produced outlined the boots, which was not a desired effect. Instead the final outputs were not rendered with mattes but layer over each other in shake. The rendering layout was highly efficient and worked well for this project.
3.6 Shader writing & texturing
The character that is the boots plays a huge role in the animation. Usually a figure would play the character and we would not concentrate on one area of the figure for any continuous amount of time. Here the boots virtually fill the frame for a considerable percentage of the animation. Being so close up means that the shader should withstand scrutiny. The boots were rendered with Pixar’s renderman, to get the best quality render, and to give the opportunity to learn and to put into practice shader writing. Shaders are written in shading language, which provides some useful built in functions such as fresnel, Illuminance, Smoothstep, and it allows fluency working with points and vectors. Since there is less programming conventions to hold you down, the idea for the shader takes precedence. Like C++, shading language is also expandable to develop your own libraries. When writing the shader all that is concerned is the colour of the surface at any given point. It was found that it was easiest to breakdown the problem into small chunks. A study of some children’s boots was produced in oil to gain a familiarity with the object, highlight the key qualities that we associate with the boots. In order to place the boots in the NPR environment the boot will have to look like it is there. This means that it is never going to look Photoreal because it could never be photographed, but the problem can still be approached as if it were realistic. It was decided that by concentrating on the main memorable qualities rather that aiming directly for realism would get the desired effect in the quickest time.
Figure 25.
Memorable Wellington boot characteristics
Highly glossy
no constant highlights but movement
fingerprints left on the boot affect the specularity
Highlights mainly consist of reflections of the surrounding environment
The sole of the boot is matte unless it is wet, in which case it is glossy
Brighter at the edges
Moulded edges, a sign that the boot has been made in two pieces
Dirt build up nearer the sole of the boot and in the small nooks
Pseudo Code
Surface
function
Calls shinyPlastic( ) // calls another function to start doing the work.
//checks to see if a colour and specular map have been defined in the interface
//if they have, it assigns them to the correct variables, which are passed to the shinyPlastic function.
Returns the final colour (Ci) and opacity (Oi), this is what is rendered for each pixel.
Color
shinyPlastic()
Calls LocIllumGlossy( )
Calls built in function Fresnel() //used so that reflections are stronger at grazing angles
Calls SampleEnvironment( )
Returns the final basecolour to the surface function.
Color
LocIllumGlossy
Calls illuminance loop()
// calculates, by looping through all of the light positions
//and brightness’, the illumination at that point
uses the Smoothstep function to create stepped highlights, rather than smooth shape defining highlights.
Returns the resultant colour to shinyPlastic function.
Color
SampleEnvironment
//Checks if either an environment or reflection map has been entered in the interface
//if it has then it calls the appropriate function
Calls Environment( ) // taken from the Advanced Renderman book,
Calls ReflMap( ) // taken from the Advanced Renderman book, used for flat objects only
(there is no need to describe
this function as it has not been used, but it does add functionality to the
program)
Returns the resultant colour to shinyPlastic function.
Color
Environment
//uses a simulated sphere
Calls raysphere( ) // taken from the Advanced Renderman book,
Float raysphere( ) // taken from the Advanced Renderman book,
Uses point E and vector I
Returns soloution
-rendering passes with Output for spec
Figure 26. In these early
shader development tests you can see the Smoothstep( ) function working,.
Figure 27. Simple UV maps
were created for the boot; both colour maps and specular maps were digitally
painted from a mixture of sources i.e. photographs, scanned in fingerprints.
Figure 28. (above) Image maps
used in the shader were painted at a very high resolution () and then had to be
converted into Renderman texture files using the existing function txmake.
(below, figure 29.) During production
the shader was being constantly tested for any irregularities, once it is
loaded into maya, it can be compiled automatically updating the shader in the
view. Renderman Artists Tools (RAT)
provides the interface which includes menus like the one below. Texture maps can accessed by the interface
provided that its variable is entered into the surface’s parameter list in the
shading language file (.sl)
Figure 29.
Figure 30. Close ups of the
boot show off the shaders ability. The
presence of an environment map is hardly noticeable on the left image, but it’s
effect can be seen more clearly on the right image.
Figure 31. Boots render
layer without environment map, left, and with, right.
[contents]
3.7 Npr
Figure 32.
In order to find the edges of the models the nizid shader is attached to the models through the use of adapters. By storing the facing ratio, (N.I) in the Red channel of the image and the Z depth in the Blue channel, when combined gives a good basis for edge detection. Shake (Apple) and Photoshop (Adobe) both provide the edge detection operation but it was found through experimentation that the Houdini (Side Fx) trace operation gave most control to create variation and so was chosen for this reason. Instead of creating a 2d image, the trace turns the image into polygon edges. These polygons can then be manipulated to create the desired effect. This gave the user more opportunity to play with the image, rather than being forced to use existing functions that create images of a very similar nature.
To give flexibility in postproduction for the user, renders were kept to black and white images that could be processed through shake to create the final image. Changes to the final look could still be made right up until the last stages.Line thickness could be decided in Houdini depending on distance away from camera. Originally it was hoped that the z-depth pass could be used to change the thickness automatically with depth, but was never realised due to the spontaneous nature of the project.
Figure 33. Early Tests
showed the technique to be possible with some work.
Figure 34. The final
procedural system in Houdini merges together different edge detection
techniques. The advantage is that the
look created is unique in style; the disadvantage is that it is hard to produce
the good results for every shot due to compromises between edge detecting the
facing ratio and the z-depth channel i.e. the closer the model is to other
objects the worse the results. This
meant that every shot had to be tweaked to get the best results.
The final NPR renders combined the use of the MANTRA renderer (renders light and shade) and the GL line renderer (renders only line images) in Houdini. After time spent experimenting it was decided that one layer renders the polygons and is used in shake to create a chalk effect. The other layer renders edges, which includes outlines. When combined together the image has interest in the detail but no depth or tonal shading.
3.8 Filming
For the shots which require the montage effect object were shot against a green screen in order to key out the undesired background. For still objects it was not necessary, but for the moving leaves and twigs it was essential.
Figure 35.
Figure
36.
[contents]
3.9 Matte Painting
Instead of modelling static geometry, it was found to be extremely effective to matte paint any arbitrary backgrounds in a digital paint package. This was made easier still by the lack of depth and tone needed, since the image construct relies on outlines to portray the images.
Figure 37. The music was
taken from the manuscript and the map was created from scratch. Images were drawn with the Illustrations of
E.H Shepard in mind. The camera move,
which tracks horizontally along this image, was created in Shake. To gain enough detail the painting had to be
a very high resolution, slowing down production, but it was still found to be
the best way of producing the shot.
In a previous project a water system using sprites was set up. Sprite animation was hand drawn. They were a few faults with the system already in place so adjustments were made to the cycling algorithm present in the sprite shader. Sprite animation was produced for frames 1-100. In the old shader the offsetting of the animation was random dependent on the particle id number.
Framecycle is the number of frames in the sequence;
imagenumber is the number image it should use for that frame;
$F is the current frame number;
Offset is a random number between 0 and framecycle;
m = imagenumber
n = ($F % framecycle) // finds the modulus, i.e. the remainder if
d = framecycle-offset
if (n =< d)
m = n+offset
else
m = n-d
Figure 38. The water system works by defining
collision objects, a source and a rough path. Models and path curves were
exported from Maya. Any camera moves
were exported with the aid of a script convchan.mel, found on a Houdini
website. The camera aperture is
different in Maya than Houdini so it also had to be converted with the help of
a simple rule found in the Houdini documents.
The camera’s never matched up exactly but the difference is minute, and
so did not cause a problem.
Shot 1 involves the stave of the music changing to form the outline of a river. This was achieved by creating curves in Maya following the required shape and then clustering control points together in order to gain control for animation. Once animated, making sure that the end points would still match up with the 2 dimensional matte painting, the curves were rendered with Renderman’s curve rendering facility. To acquire the final image quality the lines were matched to the painting in Shake.
Lighting for the scenery was void, but lighting for the boot was necessary for the shader and in order to produce shadows for a shadow pass. A basic three point light system was used to show off the shaders ability. Then new MTOR*spotlight shader was created and attached to the key (shadow casting) light. With the light colour set to black and the shadow colour set to white a shadow pass was created. These were then ready for the pipeline.
*(Maya to Renderman)
4 Compositing
A compositing build for 3 shots can be found on the
accompanying CD-ROM
Experimentation early on gave a good indication for how much time would be required for the shake project. The majority of the shots rely on a very similar tree structure. Shots that needed more 2D work such as shots 1,10,12, were allocated more time. There was also a large proportion of time allowed for experimentation. The final look had not been decided, nor was it intended to be of a particular style instead it was stumbled upon through rigorous testing.
Originally the entire project was in one Shake file, but this soon became slow running. Instead a file was created for each shot. Once each shot was rendered, a final composite file was created which essentially just edited the project together.
Figure 39. It is hoped that these diagrams visualise the Shake tree structure.
Figure 40.
The colouring for the shots compiled of a paper like background and a chalk like layer. The chalk layer adds no effect on the depth of the image and was chosen for purely aesthetic reasons.
In order to realize a Montage effect in shot10 compositing the live action material involved a less formalised routine. The objects to be composited were colour corrected to fit in with the overall colour scheme. A pass was rendered with markers positioning the objects. Since the camera is moving it was essential that this was done in order to match up the positions of the object with those that acted as collision objects in Houdini, for the water pass. Once the markers were tracked and used for the Y motion, the shot came together very quickly.
Shot 12 consists of the boot floating under the bridge. A shadow under the bridge is added simply but the boot did not look like it was convincingly moving into shadow. The glossy highlights on the filmed boot were found to be the problem. Shake Quickshapes were used to paint out each highlight but the effect was hard to achieve convincingly unless each frame was rotoscoped by hand. Instead it was found that a Lumakey node would select the highlights which then only needed to be multiplied by the grainy red colour of the boots and placed over the original boot. The final shot is subtle but effective.
5 Conclusion
The whole idea of the animated film is to suppress
the categories of normal perception; indeed its logic might even be to suppress
all differential categories and annihilate the very conditions of rationality
[Hames 95] [20]
Although these are not my words, I find this quote interesting, if a little outrageous. It is my belief that the medium of Computer Animation has a long life ahead of it, where it will develop away from the ideals of Photorealism, and challenge our normal perceptions. By mixing the stereotypes, photoreal and non-photoreal, I feel that I have suggested a new interpretation. In this way I have challenged the current trends in computer animation.
I feel that the less successful elements of the animation lie in the areas that were given least consideration from the outset. Lighting is simple and there has been little development of shadows. In hindsight I would have liked to have made the shadows softer on the boots so that the lighting seems less harsh and more in keeping with the style of the animation. Also learning and understanding shading language was a hard task and I would have liked to have innovated more in the final shader had their been more time allocated to this section.
My research of other artists helped me develop a style that combines the knowledge acquired by artists in the past, such as the Japanese woodblock print artist Hiroshige, E.H Shephard, and present such as Curtis Cassidy. I have then used these influences to create a distinct rendering style that is pleasing to the eye.
It is hoped that it is an animation that you need and want to watch more than once and one that you can enjoy both on an aesthetic level and on a more intellectual level. This is similar to the aims of the montage artist Hausmann and can be seen also to effective use in the 1998 film Pleasantville.
I would like the audience not to feel like they are pushed into thinking a certain way, there is no one right interpretation. I have realised that the animation will have a different meaning for me than others since I have established through research of memories that our past moulds our perceptions. Hopefully it will touch people and they will not quite know why.
To emphasise the theme of nostalgia I have tried incorporated the ideas used in Amelie such as using a selective colour palette. I have also chosen a universal theme of childhood to tell the story. I believe that the statements on memory made with montage have been effective and it is a theme that I hope to continue in future work.
A wide variety of techniques are used and consequently I feel that my problem solving skills have improved tremendously. I chose a challenging project which had a slight risk element. The water, which plays an important part in the narrative, was researched and developed in my Innovations Project, but there was never any guarantee that this was going to work as effectively as it did. Also going into the project without the ability to program a NPR pipeline meant that another way had to be found. I feel that the results were effective but with consequences. What would look great in one shot would look not so good in another. This meant quite a lot of tweaking ate into my contingency time.
I have always felt that quality is more important than quantity and I believe this to be true in my project. Techniques such as shader writing in Renderman have been learnt in order to get the best visual results. Most of the techniques used were versatile and gave freedom to the work. For example rendering all of the separate components on different layers meant that more experimentation with the visual style could be undertaken. Although there is not one defining strength, the final film has a unity of style that is rare to find in contemporary student work, the style is true to myself.
I have really enjoyed working on the project and feel especially proud that I have produced an animation that allows interesting interpretation and I have learnt a lot about myself in the process. It felt very satisfying watching the animation come together with the music, and I am glad that I have had the opportunity to create such a piece.
Thanks to
Steve Bell
Ben Toogood
Claudia and Lucas Moore
www.lboro.ac.uk/departments/ac/tracey/mam/woodiwiss.html
[Rosenthal 62] [1]
Erwin Rosenthal in ‘The changing concept of reality in art’ pg11
[Kelly 96] [*
Paraphrasing McLuhan 64][2] Owen Kelly in Digital Creativity
*Understanding Media by Marshal McLuhan (Routledge and kegan Paul, 1964) pg 105
[Woodiwiss] [3]
Angela Woodiwiss
[Lansdown 97] Lansdown
in Computers and Art pg12 [4]
[Rosenthal 62] Erwin
Rosenthal in ‘The changing concept of reality in art’ pg 57[5]
Herbert Zettl summarises in 'Sight Sound and Motion’ pg10 [6].
[Reynolds 02] [7] www.red3d.com/cwr/npr
[HALPER, SCHLECHTWEG, STROTHOTTE, 02] [8]
[Gooch 01] [9] Amy and Bruce Gooch pg1 of Non-Photorealistic Rendering
[Curtis 01] [10] Cassidy Curtis inside cover of Non-Photorealistic Rendering
[Rosenthal 62] [11] Erwin
Rosenthal in ‘The changing concept of reality in art’ pg 11
[Zettle 99] third edition pg 6 ‘Sight,
Sound and Motion’ [12]
[Gombrich 60] Gombrich: Art and Illusion
p169: [13]
[Giacometti] cited in Angela
Woodiwiss, Taken from David Sylvester:Looking at
Giacometti p225 [14]
[Woodiwiss] [15] Angela
Woodiwiss, Lecturer, Bedford College
The memory of familiar objects can influence what we
perceive. [Gombrich] [16] Ernst
Gombrich, cited in Arnheim,Op Cit p51
[Zettl 99] Herbert Zettl in 'Sight Sound
and Motion’ pg
58 [17]
[Delacroix 1901] [18]
Cf. Paul Signac, D’Eugene
Delacroix au Neo-Impressionism. Paris floury cited in Wassily Kandinsky,
‘Concerning the spiritual in art’ 1914 pg27
[Mackinnon 02] [19]
Ian Mackinnon ,Innovations Report 2002, Bournemouth University
[Hames 95] [20] Hames 1995 Dark Alchemy:the films of Jan Swankmajer,
Towbridge: Flicks Books. cited in “Understanding Animation” by Paul Wells
[Green et al. 99] Green, S., SALESIN, D., SCHOFIELD., HERTZMANN, A., LITWINOWICZ, P ., GOOCH, A.,CURTIS,C., AND GOOCH., B. Non-Photorealistic Rendering (SIGGRAPH’99 Course Notes). New York: ACM Press, 1999.
[Gooch et al. 98] GOOCH, A . SLOAN,P.-P.,GOOCH., B., SHIRLEY, P., AND RIESENFELD, R. “Interactive Technical Illustration.” In ACM Symposium on Interactive 3D Graphics, edited by Jessica Hodgins and James D. Foley, pp.31-38, New York: ACM Press, 1999.
[Raskar, Cohen 99] RASKAR, R., AND COHEN, M. “Image Precision Silhouette Edges.” In Proc. 1999 ACM Symposium on Interactive 3d Graphics, edited by Jessica Hodgins and James D.Foley, pp.135-140, New York: ACM Press, 1999.
[Curtis et al. 97] CURTIS,C. J.,ANDERSON, S.E., SEIMS, J. E., FLEISCHER, K. W., AND SALESIN, D.H. “Computer-generated watercolor.” In Proceedings of SIGGRAPH 97, Computer Graphics Preceedings, Annual Conference Series, edited by Turner Whitted pp. 421-430, Reading, MA: Addison- Wesley, 1997.
[Strassmann 86] STRASSMANN,
S. “Hairy brushes” Computer Graphics (Proc. SIGGRAPH 86) 20 (4):225-232 (August
1986).
[Takagi et al. 99] TAKAGI, S., NAKAJIMA, M., AND FUJISHIRO, I. “Volumetric modelling of colored pencil drawing”. In Pacific Graphics ’99 (October 1999)
7 Bibliography
‘Painting and drawing, a beginner’s guide’. By Thomas Crabb
-
‘Art of the 20th century’. Under the direction of Jean-Louis Ferrier with Yann Le Pichon
ISBN: 284 2772210
-
‘The A-Z of Art’, by Nicola Hodge and Libby Anson, Carlton Books Limited
ISBN 1 85868556 7
-
‘Drawing on the right side of the brain’ by Betty Edwards
IBSN 0-00-638114-6
-
‘Art in Motion, Animation Aesthetics’ by Maureen Furniss
IBSN 1-86462-039-0
-
‘Sight Sound and Motion, Applied Media Aesthetics’ by Herbert Zettl Third edition
ISBN 0-534-52677-2
-
‘The Oxford Dictionary of Art’ Edited by Ian Chilvers, Harold Osborn and Dennis Farr
New Edition, Oxford University Press 1988, 1997
-
‘Understanding Animation’ by Paul Wells
Published by Routledge 1998
ISBN0-415-11597-3
-
‘Animation-Genre and Authorship’ by Paul Wells
Wallflower Press 2002
ISBN 1-903364-20-5
-
‘The joy of Art’ by David Piper
Mitchell Beazley Publishers 1984 reprinted 1989
ISBN 0-600-55841-x
-
‘Japanese design- Modern Approaches to traditional elements’ volume1 e. Design Exchange
Gingko Press 2001
ISBN 1-58423-081-9
-
‘Experimental Animation- Origins of a new art’ by Robert Russet and Cecile Starr
Da Capo Press 1976
ISBN 0-306-80314-3
-
‘Concerning the Spiritual in art’ by Wassily Kandinsky
Translated with an introduction by M. T. H. Sadler
Dover Publications, Inc. New York (1914) reprinted 1977
ISBN 0-486-23411-8
-
‘The changing concept of reality in art’ by Erwin Rosenthal
George WittenBorn Inc, New York 1962
-
‘Digital Creativity’ by Owen Kelly
Calouste Gulbenkian Foundation 1996
ISBN 0 903319 73 X
-
‘Advanced Renderman’ by Anthony A.Apodaca, Larry Gritz
Morgan Kaufmann Publishers 2000
ISBN 1-55860-618-1
-
‘Non Photorealistic Rendering’ by Bruce and Amy Gooch
A K Peters, Natick, Massachusetts 2001
ISBN 1-56881-133-0
-
‘Computers and Art’ edited by Stuart Mealing
Intellect Books 1997
ISBN 1-871516-60-9
-
‘Art and Illusion’ fifth edition by E.H.Gombrich
first published 1960, Phaidon 1977
ISBN 0 7148 1756 2