Procedural Art with Computer Graphics Technology

Michel Bret

in LEONARDO, Vol 21, Num 1, pp. 3-9, 1988

Abstract—While traditional tools enable visual artists to work only on objects, the computer gives them access to the processes and sources of creative activity. Through dialogue with the model, the artist is no longer limited by the material constraints or the irreversibility of real actions. By emphasizing the process rather than the object, the computer enables formalizing of the creative act and the application of formai categories of discourse. Graduating from the hand to the brain and becoming ´intelligent´, the tool gives new meaning to the concept of artistic creation and endows the visual artist with a new attitude which leads to ´procedural art´. In order to define this concept, the author first relates the adventures which led him to change from a traditional painter to a designer of software for artists. He then describes his Anyflo System of three-dimensional animation and synthesis which uses the concept of procedural art in relation to teaching and the production of works generated by computer.


   If musicians were the first artists to use the computer, it is probably because of close links among music, language and mathematics. The visual arts, on the contrary, had traditionally had conflicting relations with formal thought and discourse. Although one can talk about painting (art criticism, semiological and linguistic study of works), a painting cannot be written and the verbal description of a painting cannot exhaust its content. However, since the appearance of advanced graphics processors and increasingly powerful computers, the spectacular progress achieved in the field of digital images has not gone unheeded by artists. Would it be reasonable to leave these marvellous image-making machines in the hands of scientists alone? Could they not, or should they not, be used for plastic art purposes? The answer to this question is not as simple as it might appear, since the computer, as a machine, is indeed a tool. But the computer is also more than that. It is what could be called a ´meta-tool´, a tool used to manufacture tools. With such an apparatus, visual artists no longer produce a work but the process which generates it. And they are no longer interested in the physical characteristics of the object but in the laws which enable the object to appear and exist.
   By simulating the creative act, i.e. manipulating a model, the visual artist can explore all its potentialities and change them at will. This attitude is impossible in real life, for one cannot go back through an irreversible sequence of actions.
   But what is left of the relationship between the artist and the material? This would seem to lead to the dematerialization and the intellectualization of art. First of all, the machine, at any given moment, can display the image of the state of the model, meaning the work in evolution. In spite of this, the spontaneity of gesture and intuition thereafter are not lost, thanks to interactive dialogue tools, such as graphics tablets and other digitizers, which provide practically instantaneous digitizing of real physical actions.


Fig. 1. Reality and simulation: diagram juxtaposing linear irreversibility and movement through network.

   A game, with its partners, aims and rules, simulates the real world, meaning human society confronted with the laws of nature. But in real life, it is death which punishes error. In the case of a model, going back through the sequence of actions and making other prior choices is not prohibited. The concept of linear irreversibility is then replaced by that of a network in which it is possible—in a given node—to select not only several aims, but also several ways of achieving the same aim (see Fig. 1).
   As such, it is possible to proceed by hypotheses or by trial and error and therefore to be working at the non-penalizing level of language instead of suffering irreversible material constraints. Far from being an inconsequential attitude remote from reality, the role of the game is essential in all learning processes. An infant plays with cubes in the same way as a pilot trains on a flight simulator—to acquire the mental structures for coping with the real world.
   Applying categories of discourse to representations of actions amounts to ´deciding´ the world before it exists; therefore it means creating it. But this could also amount to anticipating it. It is only a matter of making the model go faster than real time in order to produce an image of what reality will be. For example, the computer with its computing power and its memory plays a central role in weather forecasting and in the construction of economic or strategic models as well as in Artificial Intelligence and in all fields where human choices are decisive.
   Can one legitimately apply these considerations to artistic creation? By formalizing this process, is there not a risk of removing its intuitive and spontaneous character?
   First, there is fundamentally little difference between the attitude of a painter creating a painting and that of a mathematician discovering a theorem. Intuition alone does not explain the former, and rigour alone does not explain the latter. The discipline that an artist must observe in order to master a technique owes little to spontaneity; furthermore, it is well known that the demonstration of a theorem has nothing to do with the history of its discovery, which owes more to sensitivity than to logic.
   Second, the idea of assigning greater importance to the process than the object has been around for some time, as was clearly shown by the kinetic art experiments [1].
   Finally, my experience as a teacher has proven to me that present-day artists, far from seeing the machine as something that stands in the way of their creativity, have adopted it enthusiastically and have made magnificent use of it for plastic art purposes, as demonstrated by the in-creasing number of quality works they are producing [2].
   Like all new ideas that challenge tradition, and therefore the power of those who guarantee its continuity, computer art offends some who belong to traditional artistic circles and provokes their reticence. We should not be surprised by this. In their day, the Impressionists came up against the same sort of reaction. However, the situation is not quite the same. While Manet and Gauguin created a revolution within pictorial techniques, digital images refer to a technological universe that is far removed from the pre­occupations of many contemporary artists.
   Computer art bears the stigma of its scientific origins and, regardless of its plastic qualities (see, for example, certain ´digital-effect´ images or the films of Japanese computer artist Kawaguchi) or the sophistication of the methods used, it comes across more often as technical prowess than as genuine art. At one of SIGGRAPH´s [3] annual "Video Shows", the spectators stood up to applaud the sections they considered particularly spectacular. They were admiring the effect—and thus the programs and the machine which make it possible- and not the image as such.
   Digital images, which are derived from computer-aided design (CAD), are characterized by their ability to simulate the three-dimensional universe; this explains their frequent use in science-fiction films in which the third dimension plays a vital role. In contrast, traditional, fixed two-dimensional images refer to a less obvious and less trivial significance. Monet´s nymphéas are more than just aquatic plants. They speak the sensitive language of colours more than they replicate versions of botanists. But this fascination for representation is not aimed simply at reproducing that which exists, and, just as art has always been able to do, the new images have the capacity to create poetic universes where imaginary situations become intertwined with dream and language and do not simply describe reality.

Fig. 2. Oil on canvenas (detail) 1 m X 2 m 1971.

Fig. 3. Collage (detail), 6 m X 2.5 m, 1978.


   If computer art is still waiting for its Matisse, it is probably because artists have not yet mastered the techniques of digital images or, to be more explicit, the programming and knowledge of com-puters. At the present time, they rely too often on computer experts who play an intermediate role, with all that this involves in terms of arbitrary inter­pretation and filtering between the creator and the tool of his or her creation.
   I think it is absurd to attempt to dissociate the conceptual and practical aspects of creation. By the former I mean the project of the artist, which can be expressed by discourse, and by the latter I refer to the physical construction of the work, which is traditionally expressed by gestures and technical know-how. With the computer, this second aspect takes the form of programs and, inevitably, a certain amount of know-how.
   By extending the intellectual capacities of humans, computers, even more than the guns which extend our fists, are instruments of power, and mastering their potentialities is a social phenomenon that requires a radical change of mentality. Machines, whether we like it or not, through handling information and know­ledge (and perhaps one day even semantics), have changed the cultural landscape from literate to scientific.
   Of course, there is no need for artists to be intimidated by this new fact. As they have always been able to do, artists will use the technical tool for their own cultural ends, and it is these cultural ends which define the functional necessity of the artist.
   The two indissociable sides of any civilization, namely, the technical—that which lets artists act on the world and therefore hold their own therein—and the social—that without which the individual has no reality—are found in the functional and cultural aspects of the tool: the blade of the knife is sharp, but the design of its edge was first a written statement.
   Unlike traditional tools, which work on the material, the computer manipulates symbols. Therefore, its use in the arts involves the formalization of the act and no longer just its simple execution. This detour through simulation will enrich the action of all the power given to it by the concept of the ´model´.
   By referring to a world of repre­sentations rather than to one of objects and by speaking before acting, one can develop different strategies and make alternative choices—in short, one can bring the work in embryo into the field of social communication. This will not be simply the product of a solitary creator, but, since it is processed by language, the resuit of an exchange [4].
   The artist working on a computer uses machines that are the resuit of advanced technology (implying multiple skills), programming languages (implying compilers and Systems written elsewhere), algorithms (meaning ideas developed by others), existing programs (therefore the know-how of a programmer), data bases (programs, models, objects belonging to a common memory), etc. Far from promoting cold contact with a remote tool, the computer will bring art into social practice to an extent never before possible. Appreciation of art will no longer concern just those members of the public who are willing to make the effort to see it and to limit themselves to silent contemplation; art will penetrate every-where thanks to cable networks and be proposed as a matter to be worked and transformed rather than as simply an object to be admired.
   If the computer of tomorrow is to be a marvellous machine we can talk to in natural language and which has the power to synthesize 3-D images of perhaps even holograms in real time, then the spontaneity of gesture will be rediscovered and even largely exceeded and amplified by an ´intelligent´ machine. The artwork, in the highest sense of the word, will then be a total, interactive spectacle in which one or more spectator-creators, who will be involved in the processes both as observers and as designers, will build their own mental images in symbiosis with those programmed by the artist.
   Discussing, planning and simulating before acting certainly has always distinguished human behaviour from the more instinctive behaviour of animais. But what is new is the appearance of a tool that extends this faculty. All social activitie and therefore artistic cre­ation—will be transformed by this. What I call ´procedural thought´ is the attitude I believe will have to be adopted in regard to the computer tool.
   In the same way as designers of automobile bodywork now work on digital models, artists, by simulating their creations, can ´write´ an artwork and ´discuss´ it with the machine which, at each stage of the discourse, will give them an instantaneous image of what the work would be if it had been ´actuated´ instead of ´stated´.


   Given my scientific education and my experience with the techniques of traditional painting (Fig. 2) and collage (Fig. 3), my meeting with the computer was inevitable. This took place, purely by chance, when I joined G.A.I.V. [5] in 1976. This association of musicians, linguists, visual artists and scientists had set up a Computer Department, one of the main activities of which was artistic creation.
   For me, the computer then became a tool for artistic creation, and programming the necessary step to master it. I acquired expertise by creating various software for computer graphics and animation.
   In 1983, at the initiative of Edmond Couchot, a department section named "Image Technologie et Art" was created at the University of Paris VIII with the aim of developing ´double compétence´ in students—competence in both art and technology. During a one-year period, I had access to a VAX 11/780 computer which was connected to a RASTER TECHNOLOGY frame buffer providing more than 16 million colours. I took advantage of this to develop high-perfor-mance software which was used by students in this section and with which I produced a number of films.
   The following year, since I no longer had access to the VAX, I began using the French SM90 minicomputer. On this small computer, which was connected to a COLORIX 90 frame buffer providing 32,768 colours and video resolution, I began to develop software for philoso-phically open-minded artists. The System I call Anyflo is described below. V. ANYFLO

The Software Layers
   The software layers in the Anyflo System consist of a set of animation and synthesis programs written for artists, the structure of which is illustrated in Fig. 4. An initial layer, comprising a library of basic graphie functions, enables access to all the resources of the image processor and, in a sense, plays the parts of the brush, the palette and the tubes of paint of the traditional artist.
   But a brush without a hand to guide it is only a piece of wood. And the hand, with no consciousness to control it, is no more than an animal´s paw. Therefore, the graphic functions receive instructions from a more advanced second layer. This is an interpreter—the real heart of the System—which produces a highly inter­active interface, with a certain ´intel­ligence´, between the initial layer and the application programs.
   The function of the application programs is to stimulate the know-how and techniques without which any artist would be simply a dreamer: from synthesis to animation, including archi­tecture, mode and scenography, without excluding any future development.
   Finally, the last layer consists of ´macros´ written by the user, who can thus produce a personal work by arranging the program to suit the creative intention.

The Interpreter

Fig. 4. Diagram illustrating the software of Anyflo.

Fig. 5. Diagram illustrating the workings of the Anyflo interpreter.

   The interpreter is a loop for evaluating the character strings, which can come from the keyboard, a graphic tablet (by means of programmable ´menus´), the memory (in the form of functions written by the user) or disks (data files or executable programs) (see Fig. 5). Its originality lies in the fact that it can handle vector entities (defined as lists of ´floating´ points), which can be seen in two different ways. From the strictly digital viewpoint, all the logical and arithmetic operations can be applied globally to them, but from the geometric viewpoint, they are interpreted as 3-D polygonal lines. They can be transformed by a full range of geometric functions, and the result of this is displayed in real time.
   These objects, which are designated simply by a name, can then be entered in formal descriptions of volumes, colours or movements written in the interpreted language. The latter entities, or ´macros´, have the characteristics of functions in advanced programming languages: recursive calls, return of values, running of formai parameters, etc. While traditional practices work on objects, this type of System encourages ´procedural thought´, meaning the construction of processes whose results may be either objects or other processes.

The Synthesis Program
   As an outgrowth of the interpreter, the synthesis program uses the language to synthesize the 3-D modules and display them. The data structures are organized in recursive blocks whose significance may be, for example, to describe a polyhedral surface, colours, textures, lighting, etc. The display of objects produced as such uses the standard conical perspective as well as ´clipping´ (enabling the eye to see an object from the inside) [6].
   The lighting model used integrates the three components of light—diffused, specular and environmental [7] and also takes ´mist´ into account. Any number of light sources may be defined (variable in position, density, colour and law of illumination). An original algorithm for curved extension of polygonal lines [8] and polyhedral surfaces [9] enables the System to generate complex surfaces on the basis of a small number of control points. The concepts of modules [10] and permutational art [11] have led to the implementation of recursive patterns, which are particularly appreciated by architects (Fig. 6). A programmable alphabet enables texts to be built rapidly in space and displayed in perspective on the screen (with obvious application for credits).
   Fractal surfaces [12] were implemented according to an algorithm inspired by that of Loren Carpenter [13] to allow the simulation of mountains (Color Plate No. 3). For this process, the data base was reduced to some 30 points (which are the apexes of a 3-D grid, the shape of a horse saddle); then the process of fractalization was generated automatically by a recursive folding that imitated the shape of rocks.
   A fashion application of the Anyflo System led me to create cutting and sewing functions simulating dressmakers´ tools. And I have used ´mappings´ of natural textures to improve the realistic reproduction of certain images.

Fig. 6. Synthèse architecturale, computer-generated image, 1986. Image: Sabine Porada. Logiciel: Michel Bret.

The Animation Program

Fig. 7. Dragon recursif, computer generated image, 1979. This figure illustrates the technique of interpolation.

   The interpreted language described above can be used to assign a procedural definition to arbitrary movements. The principle consists of associating two indicators with each variable entity. One of these refers to a spatial trajectory and the other to a movement law. Through default, when there is no command, a uniform, linear animation is generated between the two extreme key frames.
   But one may intervene at any level— from the global level of the shape to the elementary level of the point to control the movements. In this way, trajectories may be associated with special points, groups of points, shapes, colours, fractalization coefficients, the eye, lights, colour tables, etc. For example, to animate a rough sea, it would be tedious to describe the movements of all the points of the surface one by one. The best method consists of writing a macro in which a simple loop on the control points alone gives them a movement in terms of their position and according to a law simulating the movement of waves.
   For choreography applications, an anchoring structure enables association of the various parts of the body of a dancer with a tree-like structure and provides automatic control of their links and coherence of their movements.

A Palette Program
   I will conclude by talking about the palette which, though a popular instru­ment, is the worst possible example of a synthesized image for the simple reason that it is not one. The palette utilizes just the computer´s speed of execution and systematically rules out its capacities of simulation and deduction, apparently in order to simplify its use. Although it represents decisive progress compared to other tools, the palette is nevertheless far removed from the real epistemological break introduced by the ´procedural thought´ concept.
   The palette program that I propose does not have this defect. It is built around the same interpreter as Anyflo and is controlled in a way that is procedural and no longer simply gestural. The spontaneity in its spatial and dynamic dimension is taken into account, but in the sign mode rather than the action mode, meaning that the image is generated by the interpretation of a gesture (through a process created by the user) rather than by the basic gesture alone.
   Thus in a classic paintbox program, the ´brush´ will follow the motions of the hand, whereas here the coordinates of points designed on the tab could be sent to a function that will do more than simply edit the ´brush´. For instance, the shape and colour of the brush could be made to depend on its speed and on the direction of drawing, on the distribution of colours in its neighbourhood, etc., according to arbitrary laws input as curves of variation or else as lists setting up vocabularies of patterns and hues.
   Figure 7 shows one example of a design obtained by such a process: curves input on the tab shape the general outline of the dragon; the ´brush´ is here replaced by a pattern interpolated between two sketches defined at the far ends of their trajectories.
   I see a double advantage in this method. As an open process—in the sense that it is programmable it does not limit the artist to the possibilities planned by the manufacturer of the machine. In addition, since the acquisition of a gesture can be done in real time—and only this gesture is significant—its inter­pretation in terms of image may be distributed through time, enabling the software to run on small, inexpensive systems.

Anyflo and Other Systems
   Anyflo is the only software that I am aware of that has been entirely written by an artist. It is thus radically different from CAD-CAM (computer-aided design/computer-aided manufacturing) Systems which, because they are built on concepts aimed at representing the world of machines, are inadequate for artistic creation. Besides, as opposed to paintbox programs, Anyflo does not rely entirely on the capabilities of the machine, but makes use of the user´s intelligence to take advantage of the tools that are proposed.
   More than a program for computer graphies, it builds up to a real language. While a System such as EUCLID demands programming knowledge, Anyflo requires only simple training of the user; effective use of Anyflo does, however, require initiative and ideas from the user. It simultaneously offers powerful graphic functions (such as realistic 3-D visuals), which can be found in other Systems like CUBICOMP, and the possibility to describe new objects and mostly new functions, which is not to be found elsewhere.


   If computer art, despite its advocates [14], has not gained recognition, this is due equally to the fact that it has not yet proved itself and to the fact that it has already disturbed the status quo too much. Artists who dare to familiarize themselves with computers at the present time—working in conditions of obscurity and facing considerable material difficulties—may well be considered, in the future, among the pioneers of con-temporary art.


1. Frank Popper, Le déclin de l´objet (Paris: Chêne, 1975).
2. I have been teaching in the "Art et Technologie de l´Image" ("Image Techno­logy and Art") (A.T.I.) section of the Fine Art Department at the Universiy of Paris VIII. A.T.I. trains artists in the field of digital images using the ´double compétence´ concept (artistic and technical). Among the works completed by A.T.I. students, one can mention the films Ballibull (1981) and Gastronomica (1984).
3. SIGGRAPH is an international show held every year in a U.S. city which presents the latest results in synthesized images. Conference reports are published in Computer Graphics.
4. Frank Popper, Art, action et participation (Paris: Klincksieck, 1980).
5. "Groupe Art et Informatique de Vincennes" (G.A.I.V.) is an association of artists and computer experts of the University of Paris VIII (formerly of Vincennes).
6. See William Newman and Robert Sproull, Principles of Interactive Computer Graphics (New York: McGraw-Hill, 1979); and James Foley and Andries Van Dam, Fundamentals of Interactive Computer Graphics (Reading, MA: Addison-Wesley, 1982).
7. James Blinn, "Models of Light Reflection for Computer Synthesized Pictures", Computer Graphics 11, No. 2, 192-198 (August 1977).
8. Michel Bret, Animation colorée sur mini­ordinateur (University of Paris VIII: Thesis, 1981).
9. Michel Bret, L´image numérique animée (University of Paris VIII: D.Sc 1984).
10. Manuel Barbadillo, "Modules, Structures and Relationships: Ideograms of Universal Rapport", Computer Art Society 12 (1970).
11. Abraham Moles, Art et ordinateur (Bruxelles: Casterman, 1971).
12. Benoit Mandelbrot, Fractals: Form, Chance and Dimension (San Francisco: Freeman, 1977), and TheFractal Geometry of Nature (San Francisco: Freeman, 1982).
13. Loren Carpenter, "Rendering of Fractal Curves and Surfaces", Computer Graphics 14, No. 3, 109 (July 1980).
14. See M. Apter, "Cybernetics and Art", Leonardo 2, 257 (1969); R. Mallary, "Computer Sculpture: Six Levels of Cybernetics", Artforum, pp. 29-35 (May 1969); and R. Leavitt, ed., Artist and Computer (Morristown, NJ: Creative Computing, Harmony Press, 1976). Contributions by R. Mallary, E. Zajec and G. Csuri.


Artificial Intelligence—this name covers tech­niques which enable computers to simulate human behaviour involving deduction. One example is the ´expert System´ which has a ´knowledge base´ (equivalent to the knowledge a human being stores in his or her memory throughout life) and an ´inference motor´ (equivalent to the deductive capacities of the human brain).

compiler—a program that translates a source program (written in advanced language) into the binary language of the computer.

floating point—mixed numbers, i.e. numbers coded with a decimal parity, as opposed to whole numbers.

graphic processor—an apparatus that generally is connected to a computer, from which it receives instructions enabling it to store a digital image, apply a certain number of processing operations to it and display it.

graphic tablet or digitizer—an apparatus comprising a flat surface and a pen, the positions of which are sent to the computer to which it is connected. There are also three-dimensional digitizers.

menu—simulation, in the form of a table displayed on the graphie screen, of a function keyboard. Execution of an associated procedure is provoked by selecting a box.

pixel—the smallest element of the screen of a graphic processor, corresponding to one word of the memory in which its colour is coded.

Fig. 8. Paysage fractal, computer-generated image, 1986. This figure illustrates the technique of simulation of mountains through use of fractals.