Evolutionary art

Evolutionary art is a form of generative art in which works of art from areas of the visual arts , music and also the performing arts are created using evolutionary algorithms . Evolutionary algorithms are methods of solving optimization problems using principles of natural evolution . By understanding artistic processes as optimization, objects can be created that have an aesthetic effect on people . For reasons of time, this class of algorithms must be implemented with the computer , but could in principle also be calculated by hand. Evolutionary art therefore belongs to digital art .

Basics

As with all evolutionary algorithms, evolutionary art is based on a population of individuals, each representing a visual structure. This representation can either take place indirectly, in that, as in genetic programming , individuals each contain a program that creates a visual structure so that the biological distinction between genotype and phenotype is maintained here. However, the representation can also be direct, as in the case of evolutionary strategy , in which an individual is viewed only as a phenotype to which evolutionary operations are applied. In this case, an individual contains an image, drawing, moving image or the like in the sense of an image file or video file .

Art process

Virtually all of the evolutionary art applications that use indirect representations produce non-representational visual works. Regardless of whether direct or indirect representation, there are only a few approaches to objective evolutionary art.

In the evolutionary art process, a starting population of individuals is first determined. In the case of an indirect representation - as is common in genetic programming - random programs and thus random visual structures are generated. With a direct representation, mostly non-random visual structures are selected by the artist, e.g. B. Images from previous evolutionary runs.

A reproductive phase follows, in which the present individuals are reproduced in accordance with a reproductive strategy by applying recombination and mutation operations to the representation structures. The type of these operations depends on the type of programs or direct visual structures, just as in evolutionary algorithms, for example, linear and hierarchical individual structures generally require adapted recombination and mutation operations.

Part of the reproduction strategy is the way in which individuals are selected for recombination (selection for reproduction). If the reproductive strategy is based on genetic algorithms , fitness values must first be available for each individual. The frequency of selection for reproduction is a strictly monotonous function of this fitness; H. the higher the fitness, the greater the likelihood of choice. If the reproductive strategy is based on evolutionary strategies, the selection is randomly distributed.

After the reproductive phase, a population of offspring results, for each of which a fitness value must be determined, which in some way should reflect the aesthetics of the visual structures. An algorithmic determination of these values ​​would require a formal aesthetic model, which in previous processes for evolutionary art does not exist or is only partially available. Therefore, algorithmic methods are limited to the determination of simple properties of the image analysis and models based on them, such as B. entropy-based models. The determination of fitness by a person or a group of people (interactive evolution) is widespread. Usually this is the artist who determines the evaluations according to his subjective aesthetic criteria. Alternative methods for empirically estimating fitness are, for example, the time that a viewer looks at a visual structure that is presented to him. There are also preconscious methods in which an attempt is made to derive a correlation between physiologically measurable properties of a viewer and his or her aesthetic assessments (e.g. pupil reactions ). The most innovative approaches here are neuroesthetics , in which brain regions are identified that are involved in aesthetic evaluations and in which correlations are to be established between the activities of these regions and aesthetic evaluations (analogous processes such as neuromarketing ). However, since these approaches require complex and still very expensive devices for medical imaging , their use in evolutionary art has so far been limited to isolated small studies.

If parents and offspring each have a fitness value, a selection strategy is used to determine which individual will be allowed to continue to exist in the next generation and possibly reproduce. This selection strategy takes into account either only the offspring or the union of parents and offspring. If there is no further termination criterion, such as reaching a predetermined maximum number of generations, the next iteration of the evolutionary art process is started with a new phase of reproduction.

Non-photorealistic rendering

One application of evolutionary art is non-photorealistic rendering , an area of computer graphics in which graphics are deliberately not displayed true to their physical image. One example is the generation of an artificial painting from a photograph . In 2005, British scientists Collomosse and Hall developed an algorithm that creates paintings from photographs. A painting is understood as a sequence of brushstrokes, with brushstrokes being defined by attributes such as position, direction, color, etc. A genetic algorithm is used to search the space of all possible paintings in this way. The fitness function , which assigns a quality to each solution candidate , compares the edge image of a candidate with a salience image calculated at the beginning . The salience of an image detail indicates how conspicuous it is to a human observer. In Collomosse and Hall's algorithm, the salience of the image details is made up of three factors: rarity, visibility and a third factor that first learns the taste of users in certain areas in order to distinguish artifacts that are important to humans from unimportant ones.

The salience calculation is based on the idea that works of art “do not represent a mirror” ( Ernst Gombrich ) of reality, but rather an interpretation by the artist.

See also

• Evolutionary aesthetics

literature

• Peter J. Bentley (Editor): Evolutionary Design by Computers . Morgan Kaufmann Publishers, 1999, ISBN 978-1-55860-605-0 .

• Philip F. Hingston, Luigi C. Barone, Zbigniew Michalewicz (editors): Design by Evolution: Advances in Evolutionary Design . Springer, 2008, ISBN 978-3-540-74109-1 .

• Juan Romero, Penousal Machado (Editor): The Art of Artificial Evolution: A Handbook on Evolutionary Art and Music . Springer, 2007, ISBN 978-3-540-72876-4 .

• Stephen Todd, William Latham: Evolutionary Art and Computers . Academic Press Inc, 1992, ISBN 978-0-12-437185-9 .

• Mitchell Whitelaw: Metacreations: Art and Artificial Life . MIT Press, 2004, ISBN 978-0-262-23234-0 .

• Karl Gerbel, Peter Weibel (editor): Ars Electronica 1993: Genetic art - artificial life = Genetic art - artificial life . PVS-Verleger, 1993, ISBN 978-3-901196-07-2 .

Web links

• Thomas Dreher : History of Computer Art, chap. IV.3: Evolutionary Art
• International Conference on Evolutionary and Biologically Inspired Music, Sound, Art and Design
• MediaWiki-based encyclopedia evolutionary art with the most comprehensive online bibliography for Evolutionary Art and neighboring areas such as evolutionary architecture, design and image processing, generative art, Computational Aesthetics and Computational Creativity

Individual evidence

1. ↑ http://srooke.com/
2. ↑ Archived copy ( Memento of the original dated May 30, 2013 in the Internet Archive ) Info: The archive link was inserted automatically and has not yet been checked. Please check the original and archive link according to the instructions and then remove this notice. 
3. ↑ JP Collomosse and PM Hall: Genetic Paint: A Search for Salient Paintings , 2005
4. ^ Ernst Gombrich: Art and Illusion , Phaidon Press, Oxford 1960