Past images of the future
Arts et Technologies de l’Image
Université Paris 8
Abstract. The majority of the techniques known until today make it possible to create images which solidify one moment (painted, photographed, filmed, etc.) or which is held at the present (interactive installations, performances, 3d in real time). The creation of artificial worlds, having their own clock of time makes it possible to explore the future finally, with various stratagems. The use of the methods of evolutionary creation is used to travel in the virtual time of worlds of artificial beings. Moreover, when these worlds are connected to the real world (by images, photographs or video camera in real time), one can plan to visualize our future, seen by the evolutionary creatures of another world.
What the images coming from painting, of photography and the cinema, to quote only the most known techniques, have in common? They all are of the images which are fixed on supports, and which have the characteristic to describe last moments. Painting solidified a still life or a landscape, the photographer captured an image of an animal in action, from one strong moment of sport, the scenario writer assembled a film with catches of sights turned in the previous months. All these moments are passed (within the meaning of our time, such as we live it as human). We will keep ourselves of course statement which they are techniques of the past (not to confuse...). But, they produce images of past (more or less near). Even a photograph taken on a numerical camera that one can visualize almost instantaneously, shows one last moment of a few seconds.
1 From past to present
Of course, many scenario writers of films of science fiction propose to us, which are supposed to be held in 100 years, thousand years, etc. Some is the date on which the action of film proceeds, we look at images which are fixed on a film: they are thus recorded images, in the previous months for the diffusion of film, even if the subject is futuristic: their contents are future, but their form, their structure even passed. However, of many artists already tried to play with time. Among the most famous examples, it is significant to quote the movement cubist, and certain famous fabrics of Picasso, where one sees on the same painting several sights of the same character, at various moments of its action. In another style, Marcel Duchamp endeavours to represent on only one painting of the figures moving, like his “nu descendant l’escalier” (see figure 1). The current most extremely in this direction relates to the Italian futurism, which tests many techniques to represent time in photographs, paintings and other works of art.
Figure 1 : Nu descendant un escalier. Marcel Duchamp
But it will be necessary to await new ways of creation and mainly the arrival of the computer to generate images in real time, through interactive performances or works to see appearing works of the present. Of course the painter whom one looks at carrying out his work is also "in real time", just as the artists who use it video in a direct way. It is with the advent of the 3d real time, and the virtual ballades in the medium of fictitious worlds that the production of images at the present takes really its rise: it is the era of the interactivity in all its splendour, where the witness can interact on the work of the artist, to walk there, but even if required to modify of them certain aspects, etc.
2 Artificial life and its clocks
And then the first works related to the artificial life made their appearance. Generally using evolutionary techniques to make evolve/move creatures in their environments, these works have the characteristic to put to us in front of new worlds, populated by strange creatures, which have some kind of life. Among the strong currents having explored the first this type of technique, it should be quoted the artists that us (like Mitchel Whitelaw [ 16 ] says it well) the “breeders of art” will name, or “breeders of images” (or of sounds, etc). Among the famous artists of this current, it is necessary for us to quote names like William Latham [ 1 ], Karl Sims [ 2 ], Steven Rooke [ 3 ], Penousal Machado, Jeffrey Ventrella, Driessens and Verstappen [ 7 ], etc (see figure 2). These artists use the genetics to make evolve/move worlds of images and creations. They are the precursors of the artists who straightforwardly will create worlds of artificial life, letting evolve/move their works with the wire of time, the time of a performance, an exposure, or even uninterrupted by using Internet network. We could study many characteristics on the creation of such worlds and creatures, but one of it was often neglected: the aspect of time. Indeed, dice the moment when these creatures are generated, it is held a certain time, which is measurable in general by a clock in the computer. In theory, this clock is supposed to follow that of our time, which one will name "human time", but does anything of it does not oblige there. The clock of the computer has this characteristic, which it can be even modified by us or a data-processing program who has any capacity to advance time, to move back it, slow down it or accelerate it. We can with ease travel in the time of our computers, simply by changing the date of the system or by accelerating the rate of the clock: the future of the virtual worlds that we create is thus (enough) easily visible! The creations based on the artificial life thus have this characteristic to be images, which can belong to their own future, and it is undoubtedly the first time that that occurs in the history of Art...
Figure 2: Genetically advanced self-portrait in the space of Mandelbrot. © Ventrella. 2005
3 To play with time
It is not quite difficult to understand that while advancing the clock of a computer, one can go "to see" an image of the future (if one considers the time measured by the computer as the principal scale of measurement of time for beings which exist only in the computer). For example, an image which would consist in plotting a curve of a F(t) function where t represents the life time of the image (for example: f(x) = 2*x), can be seen with its starting point, in real time at the time present, but also, suddenly and suddenly in its future, simply while advancing artificially t, measurement of the internal time of the computer. Not really very revolutionary, certainly. But, this becomes already definitely more interesting when one uses the techniques of evolutionary programming. If it is supposed that one makes evolve/move an artistic work on the basis of genetic algorithm, one in the same way can, outward journey to see what will occur for the future generations from these worlds, and to go to explore the evolutions of these worlds in thousand years or more. Of course, it is necessary for that to have a certain computing power: but nothing stops a technician who can, if it wants of it, simply use network computers for example, calculate twice more quickly an evolution of virtual world (with 2 computers) or 20 times or why a million times, with a coupled use with Internet network ! This concept is not easily conceivable for our human brain, but after all, it is not more astonishing than the concepts of curve of the space time [ 17 ] which controls us: also we decided to explore evolutionary creations of the future, this in order to seek to open new tracks to discover how to produce images of the future.
4 Evolutionary creations
Most of the time, the artists or the scientists who create virtual worlds, pass most of their time to explore and study these worlds, with our own clock of time, and also our own knowledge of the world. We endeavour here to make the reverse exactly! Indeed, since we create virtual beings, which have a certain degree of intelligence (at least of behavioral action) and a certain way of behaving in their virtual world, it appears interesting to see our quite real world, since the virtual world, which we created. We will take again here the world of the "beings paintings", described by the author in particular on the last Siggraph (see figure 3). They are beings whose body is composed of virtual pigments and whose principal occupation is to organize itself in a 3D space to generate images. They are thus able to be driven, and to colour multiple ways, according to various parameters, according to evolutionary methods. In preceding work, especially "we were to like their primary education goal", because the artist could then visually choose parts of the world of the "beings paintings" which it likes and inspire it. But here, we propose to study the reactions of these beings in front of series of images that we propose to them: photographs, drawings, paintings of large Masters and even of the real image time in video with the use of Webcam, quite simply. The method of creation will be found extremely moved: we do not look at to only any more "being them paintings" to organize itself to create images. We will try to see how virtual beings can react to our world, how they can create, by looking at us.
Figure 3: Vision on the world of the “beings paintings". © Alain Lioret
5 Attention, these images look at us!
The choice "to see" the world since that of the virtual beings that we create is far from being pain-killer. Indeed, it will be necessary in this case to try to include/understand how these beings there can perceive our world. And a priori, they are well far from perceiving it as one can think it. They do not have for example any notion of the objects of our environment: they do not know what is a chair, a body, a face nor even a flower. When one presents an image or a video to them, they "see" just a juxtaposition of pixels, an arrangement of colors, whose they will be able to analyze the composition. Indeed, these beings are data-processing creatures. They cannot know a priori what an image with our direction represents: on the other hand, they can try "to include/understand" an image with the tools they have. It is there that the concept becomes interesting: the only tools, which one can allot to this type of beings are generative tools, functioning on the basis of parameter. It can be a question of software of creation of fractals, cellular machines, L-systems, data-processing languages (3d for example), of tools for generation of images by genetic algorithms, networks of neurons, filters of images, etc. Thanks to this type of tools, the "beings paintings" can try "to include/understand" the images that one proposes to them by analyzing them, and while trying to reproduce them more or less perfectly. This stage, they finally do not seem more advanced than owe us the great mysteries of the universe. But, they can try to analyze our images by using genetic algorithms to produce data-processing generations of images, whose aptitude can be measured by a resemblance to what they "see". (see figure 4).
Figure 4: an image analyzed with fractals and 2D filters by the beings
© Alain Lioret
6 Genetic analysis of images
Our "beings paintings" can thus use all the software, which they want to try to analyze our world. Dice which they are tools being able to be launched automatically. This principle is very open besides, and one can provide them a list of tools which they can use. Here is a small example:
- Ultra Fractal
- Visions of Chaos
- Filtres Gimp
- Apophysis 
- Kandid (voir figure 5)
- MEL sous Maya
- Mathematica 
As one can see it, this small list example is very open, and non exhaustive of course. In fact, dice that a software able to produce image has an integrated language, which makes it possible to describe a process of creation of image, it can be used in the total diagram of analysis of image of the virtual beings. One in the only ways for these beings of trying "to include/understand" our world is to compare their creations with our images, pixels by pixels (they do not have a priori a sense of organization in lines, circles, etc, but they can "learn" that if one educates them in this direction). It should be noted that this method takes as a starting point the work of Jeffrey Ventrella, which used this kind of technique to create self-portraits in space fractal of Mandelbrot. The aptitude of the virtual beings to include/understand our world can thus be simply measured by adding the errors on each pixel with an image that one presents to them. Fitness = S1, N e(xy)
Where e(xy) represents the difference in color of the pixel of coordinates xy between the real image and the image generated by the "beings paintings".
Figure 5: use of the Kandid tool to analyze the human images by the virtual beings.
7 Images, films and works in real time
The general outline of creation of these beings is simple: one or more images of our world are presented to them (photography, video, painting, etc). They then will try to produce populations of judicious generative images to resemble the model suggested: in that, they are comparable with our artists who will try to reproduce a landscape or a still life on a fabric: but their tools are very different. With genetic algorithms which measure the differences in image, they try to reproduce their model as well as possible. This stage, it is significant to distinguish two types of very different tools. The tools which generate an image solidified in time (like a raytracing in PovRay language) and the tools which generate an evolutionary image according to rules' (cellular L-systems or especially automatas). The first category of tools is often useful to implement to manage to produce more realistic images, but hardly allows to play with the internal clocks of the creatures. The second category on the other hand is much more interesting, since by providing rules of construction of the images, which are closely related to the process of generation in time, it makes it possible to make evolve/move the images according to their "age", of their birth to their future. It that one can plan to study "our future", is seen here by the beings of a virtual world. Let us take a simple example, (drawn from Mcell, software of cellular automatas [ 8 ]). Let us suppose that a "being painting" produces image 6. This image is the resultant of a process of a cellular automata visualized at one moment T, starting from complex rules. These rules can thus make it possible to generate the future of this image, simply continuously the generation of the cellular automata.
Figure 6: a creation at the moment T by a cellular automata.
This very general process of creation will enable us to also create fixed images of course, but also some works in real time, perpetual genetic evolution, and even of quite simply astonishing films, bus to generate by virtual beings having their own design of the world. We thus carried out the film "Galateia 2005" with this methodology, while presenting at a group of "beings paintings" of the series of fixed images, which were used here as storyboard for the creation of film. The realization of the images is very complex here, and includes/understands several layers of images, realized with very diverse tools (population of the drawn software tools randomly). The movements themselves of pixels from one key image to another were analyzed according to same process's by the virtual beings which translated and analyzed movements of pixels produced by a parallel video capture. Moreover, each time that a cellular automata was implemented in the design of a key image, the preceding and following images (last and future) was calculated. Galateia thus integrates images of the future. This experimental film is a good example of evolutionary creation.
8 The first experiments of creations of images of the future
It will be included/understood well that the evolutionary techniques of research of creation of images are interesting in more than one way. However, our attention is held definitely here by the possibility of exploring the future, our future even, and this for the first time in the history of art. Thus, the use of WebCam in real time can make it possible the "beings paintings" to collect our world on line, to analyze it with their way, with their tools and to deduce their evolutionary rules from them. These rules make it possible to generate images close to what one can show them, and to deduce from it their past and especially their future! And it is indeed our future, seen by virtual artists, but of our future nevertheless.
9 About the beings paintings
The "beings paintings" themselves can be created multiple ways. We mainly used the simulator of Artificial Life Framsticks [ 9 ] to generate the beings paintings, but we also made tests with other environments, in particular Breve. As we explained in preceding work [ 11,12,13 ], our beings paintings have only one objective in the Life: to organize itself to produce images. Previously, the produced images were chosen by the user. With this new work, the images carried out try to resemble the models that one proposes to them. The visual analogies can be done at various levels, color or with a simplification black and white. One can even use certain filters (for example detection of contours) to facilitate the work of "imitation" [ 10 ]. Each being painting belonging to the population of the creative beings thus will analyze us with his way, with the generation of populations of solutions, which will utilize various tools 2d or 3d, free or commercial, according to those which are available. It is thus simulated genetics on double level, which is implemented, with evolutionary populations of creatures painters which generate solutions of combinations of process of creation of images.
Figure 7 : A creation of the "beings paintings". © Alain Lioret. 2005. Extract of Galateia
10 Applications and perspectives
The techniques implemented here can be compared with reverse engineering. In addition to the artistic creation seen under the eye of other worlds (but also with their systems of thoughts, their means and tools), one can make use of it for many graphic applications. Indeed, as one explained, the method works with a large variety of tools (those having a language, methods of scripting or equivalent) and allows artists or technicians of the image to seek to generate images according to a certain configuration, with tools not intended a priori for that. The centers of interests are rather significant, because they must allow the creation of new fashions of returned images (2d and 3d), and involve the appearance of libraries of 2D filters, 3D shaders , procedural textures, etc. The installation of this type of technique in the form of plug-ins in great software must make it possible to the creators to make emerge new methods of creations, and to set up of the extremely rich but complex processes creative, so complex that it would have been impossible to implement them manually nor to even think of it...
1. Latham, William. The Conquest of Form: Computer Art by William Latham. Arnolfini Gallery, Bristol, December 3rd 1988 - January 15th 1989
2. Sims, Karl. Evolving Virtual Creatures. Computer Graphics (Siggraph '94 Proceedings), July 1994, pp.15-22
3. Rooke, Steven. An Introduction to Creative Evolutionary Systems. In Creative Evolutionary Systems, p339-365. Editions Morgan Kauffman. 2001
4. Ventrella, Jeffrey. Self portraits in fractal space. 2005
5. Wolfram, Stephen. A New Kind of Science. Editions Wolfram Media. 2002
6. Draves, Scott.(Apophysis) Site : http://draves.org/art.html
7. Driessens, Erwin et Verstappen, Maria. Site : http://www.xs4all.nl/~notnot/
8. Wojtowicz, Mirek. (MCELL). Site : http://www.mirwoj.opus.chelm.pl/ca/
9. Komosinski, M. The World of Framsticks: Simulation, Evolution, Interaction. In: Proceedings of 2nd International Conference on Virtual Worlds (VW2000), Paris, France, July 2000. Springer-Verlag (LNAI 1834), 214-224.
10.Hertzman, Aaron. Painterly Rendering for Video and Interaction. NPAR. 2000
11.Lioret, Alain. Emergence de Nouvelles Esthétiques du Mouvement. Editions L’Harmattan. 2004.
12.Lioret Alain, Emergence of L-cellular morphogenesis. Generative Art 2004.
13.Lioret, Alain. Beings Paintings. Art Gallery. Siggraph 2005
14.Capra, Fritjof. La Toile de la Vie. : Une nouvelle interprétation scientifique des systèmes vivants. Editions du Rocher. 2003.
15.Flake, G.W. The Computational Beauty of Nature. MIT Press. 1998.
16.Whitelaw Mitchel. MetaCreation. Art and Artificial Life . MIT Press. 2004
17.Greene Brian. The elegant universe. Ed. Robert Laffont. 1999.
© Alain Lioret. 2005. Extract of Galateia.