Soka Univ.> Eng. Dept.> Tatsuo Unemi> SBArt> SBArt4


by Tatsuo Unemi, Soka University. January 2010 - July 2011.

Click here to see the demo video.
This is a project of an automatic art in which the computer autonomously produces an infinite sequence of audio visual pattern. A custom software developed by the author is taking a main role of the work, that based on a genetic algorithm utilizing computational aesthetic measures as fitness function. Starting from a population randomly initialized with mathematical expressions that determines the color value for each pixel in a rectangular area, a never-ending series of abstract animations are continuously displayed on the screen with synchronized sound effect. The visitors will notice not only the recent progress of the power of computer technology but also will possibly be given an occasion to think what the artistic creativity is.

Keywords: screen-based installation, generative animation, autonomous evolutionary system.


Human’s recognition system is always seeking something interesting, because it is often useful for his/her life. Beauty and ugliness of living systems are also the target of attention that have emerged through the evolutionary process for a billion years in the ecological relations between recognition system and morphology. The recent improvement of computer technology and the achievements of Artificial Life researches are providing a number of feasible methods to cause complex phenomena on computer-based artificial system. Visualization of such phenomena often amazes us. It is an alternative method of art to find some piece in an artificial world instead to produce something by human’s hands.

Interactive evolutionary computation is one of such approaches in a style of interaction between computer and human. The computer generates a number of offspring as candidates that hopefully acceptable by the user, and the user judges his/her subjective evaluation for candidates that should remain as parents for next generation. The iteration of this process easily leads us to a masterpiece that is too much complex to design by human’s hand, in a similar way that human beings have been applying breeding to almost all of agricultural products and pets.

On the other hand, recent researches on computational aesthetics have been revealing that it is not impossible to calculate interestingness on images. Utilizing the power of parallel processing for graphics in the latest personal computers, it is possible to realize a system that produces never-ending series of live animations through an evolutionary process based on computational aesthetic measures as fitness function. It shows the visitors newly evolved abstract animation each time that is also new for the developer because the process is fully autonomous and generative.

These technologies are useful not only to build up a system that makes unpredictable interesting phenomena but also to provide an occasion for people to reconsider how we should relate to the artifacts around us. We know the nature is complex and often unpredictable, but we, people in the modern democratic society, intend to assume that artificial systems should be under our control and there must be some person who takes responsibility on the effects. The author hopes the visitors will notice that it is difficult to keep some of the complex artifacts under our control, and will learn how we should live with them and how we can enjoy with them.


Technical Features

Evolutionary computation constitutes one of the most powerful mechanisms to create complex phenomena that are hardly restricted by the author’s own subjective preferences. This work is based on a custom written software entitled SBArt4 version 3 that has been developed by the author. Similar to the evolution of biological organisms, the software maintains a population of individual animations, each of which is subject to generational changes via the mechanisms of reproduction and selection. The genome of an individual animation consists of a mathematical expression that serves to calculate color values for the pixels in each frame based on the pixel's x and y coordinates and the frame's temporal position. New animations are born by combining the genomes of parent animations via crossover and mutation. The selection process takes into account sequences of frames and attributes higher fitness values for animations that better meet a set of quantitative aesthetic criteria. These criteria are derived from image processing techniques and deal with measures for information complexity, global contrast, and statistical color distribution. These measures of computational aesthetics play an important role to automatically evaluate digital photographs and drawings. Accordingly, they constitute very promising ingredients for fitness functions in evolutionary forms of art.

In this work, the process of evolution and the playback of the animations happen in parallel. Those animations that possess a high relative fitness value are shown one after the other for 20 seconds each. The phylogenetic tree as a trace of the animation’s ancestors is super-imposed on the animation to aid the visitors in understanding how the animation was born. The animation is accompanied by synchronized sound effects that are automatically synthesized from statistical features of the animation’s frames.

This is a screen-based installation that projects a type of abstract animation automatically generated by custom software running on one or two personal computers, MacOS X 10.6 or 10.7. Two computers are better to make a smooth animation by separating computations for generation and rendering. It is ideal to setup a large screen and hi-quality speaker system in a dark room, but smaller display of HD resolution is also acceptable.

Visitor’s experience

This installation has no interaction with the environment. The visitors can enjoy both a never-ending animation on the screen and sound effect from the stereo speaker system. The variations are organized in a generative way using a evolutionary computation so that the visitor will have a new experience at any moment. Some explanatory images, such as a phylogenetic tree, are also super-imposed over the generative abstract animation.

Created in July 26, 2011, revised in March 9, 2012 by T. Unemi.