Nature promotes the evolution of organisms by means of the reproduction, mutation and natural selection. The information that determines the features or characteristics of each organism resides in special entities called genes. These genes are in the DNA of each cell. When two organisms are sexually reproduced their DNAs are combined and their child is the result of the mixture of their parents’ characteristics. In this way, nature renews the species by obtaining new characteristics. The changes obtained from a reproduction may represent either an advantage or a disadvantage. The only way of knowing this is to observe if the organism adapts itself to its environment. The organisms that cannot completely adapt themselves to a particular environment could do so in a different environment; therefore, the best criterion for evaluating this is their survival. Those organisms that cannot adapt to an environment die, while those that are well-adapted are reproduced. Therefore, only the ones that can adapt themselves would be able to legate their features to the other generations. In this way, the evolution produced from generation to generation is of adaptation. Apart from reproducing, nature obtains changes from the mutations. A mutation is a variation at random produced in the DNA chain. This variation affects the genes of an organism and, in consequence, those of their sons as well. Most of the times the mutations are harmful, but when they represent a benefit to the organism, they make it more adaptable and the probabilities of reproduction are higher and the benefit is legated to their sons.

In computer science (Artificial Intelligence), it exists a technique called "Genetic Algorithms". This technique imitates the mechanisms of genetic evolution and applies them for solving problems. The steps of this algorithm are the following:

I) For each problem a set of solutions are generated, each of them representing an individual. As these solutions are generated at random, they are seldom the best ones, but there are some better than others.

II) As it happens with individuals, solutions are set in pairs and are reproduced. From each couple, son-solutions are obtained with a combination of their parents’ characteristics. In this way, one can obtain a new and better solution from the mixture of two different solutions.

III) Depending on an adaptation function that evaluates how efficient the solution is for solving the problem, a ranking of each solution is prepared. The worst solutions of the ranking are eliminated and the best ones are chosen. This imitates the process of natural selection.

IV) Here again solutions are set in pairs and they are reproduced. This process is repeated until obtaining a satisfactory level of solutions. The idea of the algorithm is to combine the solutions to obtain some better ones sporadically. Only those that represent an improvement to the population are selected to legate their characteristics when combined with others. It is important to know that the algorithm produces mutations in the genes at random with the aim of generating a variety that does not depend on the population itself.

Our idea was to apply genetic algorithms to an art work so as to obtain improvements on it. The first problem that arises is to decide what constitutes an improvement on an art work. Is it possible to face the production of an art work in the same way we face the solution of a problem? Does it exist an objective criterion for evaluating and improve an art work? We believe it does not exist such a criterion. From our view point, the evaluation of an art work is subjective and at most conventional. However, that is not our central problem. Though the subjectivity of the criterion for evaluation, a genetic algorithm that improves the work according to the user’s choice could be constructed. But the real problem is that it is almost impossible to model someone’s choice. In other words, the user can know which choice is better for him, he can also, though with more difficulty, explain which criterion he is using, but he would find it impossible to translate this criterion into mathematic or algorithmic terms to create a function. However, we found a solution to this problem: we leave the user to evaluate freely, instead of creating a model for evaluation. We proceed in the following way: we put an art work in the world wide web (belonging to the same specie), we allow the user (spectator) to go over the collection and to select two works. This selection has a double function: on the one hand it sets the couple to be reproduced, and on the other it increases the ranking of the works that were chosen. Therefore, it is possible to know how many times each work has been visited and how many of these times it has been selected. The most selected works will survive. Under this criterion, the works will evolve in relation to the choices of the population that connects to the site or at least in relation to the majority of this population.

Victor Vasarely was an optic and kinetic artist. He was a designer, a painter and a sculptor. He was born in Pecs (Hungary) in 1908. In France he worked as a publicist until 1936 when he started to develop his graphic work. In 1952 he definitely opted for the abstract and he became popular for having developed the Op-art (Optical art). Optical art is a style based on the changing character of a work due to the optical effects. Vasarely considered the shape and color as the structural elements of his work. Something characteristic of his artistic orientation is the break of the plane character of the square. He created the optical illusions of depth and movement by skillfully disposing simple figures such as circles, squares, cubes, ellipses and hexagons. Vasarely has been considered one of the most interesting and revealing artists. He was also a good theoretician who preferred to be called artisan rather than artist. His will was to create a synthesis of painting and plastic arts, in which the painting has all the prerogatives of a two dimensional sculpture. He did not believe in the existence of a unique and singular work and he believed in the democratic nature of art. When we first saw Vasarely’s works, we immediately notice that they had a modular structure and that the variation of this modules followed a strict logic of composition, almost mathematic. The simplicity of the modules and the power of the structure that Vasarely created for articulating them decided us to choose him without doubting. Vasarely’s works follow a constructive plan, a sequential organization of operations applied to an initial configuration. It can be said that his paintings follow an algorithm. We consider that this modular and algorithmic structure is ideal to be represented and operated by genetic algorithms. We found Vasarely’s work ideal for carrying out our project. It was even a greater surprise for us when we got to know the democratizing intention of the artist, which fully coincided with our intention to make the spectator participate in the creative process. It is important to say that the fact that Vasarely’s work was so useful for our ends is not a casual coincidence, rather, it is because our final intention is the same as his. Therefore, we believe we are just prolonging his ideas.

One of the most important stages in the design of the reproduction engine (the software in charge of the genetic reproduction of the chosen paintings) is obtaining a good genetic reproduction of the paintings. That is to say, to analyze and to extract from the paintings their constitutive features so as to be able to separate them into genes. A good genetic representation is one in which the combination of their parents’ genes produces a child that inherits their characteristics in a noticeable way. If the features of a painting are not properly represented in the genes, they will not be inherited. To make the genetic representation we analyzed a set of paintings that had the same sort of characteristics. These paintings can be described in the following way: a matrix of their constitutive features so as to be able to separate them into genes. A good genetic representation is one in which the combination of their parents’ genes produces a child that inherits their characteristics in features of a painting are not properly features of a painting are not properly represented in the genes, they will not be inherited. To make the genetic representation we analyzed a set of paintings that had the same sort of characteristics. These paintings features of a painting are not properly represented in the genes, they will not be inherited. To make the genetic representation we analyzed a set of paintings that had the same sort of characterics degradee’s zones: they divide the canvas. Changes of the parameters of the canvas associated with each zone.

The genes we chose to represent these characteristics of the paintings are:

Gene 1: Size of the canvas and zones.

Gene 2: Kind and direction of the degradee.

Gene 3: Changes produced in the zones.

Gene 4: Initial figures of the canvas.

Gene 5: Palette.

Once the genes have been defined can we make the reproduction between two paintings. When the user selects two paintings for their reproduction the engine takes a gene of each of them at random until completing the DNA of the new painting.

(Biopus Project participated in the Universidad Nacional de La Plata Expo that took place

at Cine Select in Centro Cultural Pasaje Dardo Rocha on the 30th August 2002.)