By simply pressing the space bar of his laptop prof. dr. ir. Loe Feijs starts his program. On the screen we see a brother (or sister) of the Victorie Boogie Woogie arise within thirty seconds. Starting from randomly chosen ‘cell nuclei’ the characteristic yellow lines with the colored small blocks first emerge, which determine the rhythm of the Boogie Woogie. Subsequently colored planes appear between these lines, which are in turn filled in with smaller rectangles.
Feijs opens a window with the code and adjusts a parameter. “This produces a composition with some more detail. You’ll see that the computer also takes a little longer to draw this.” He has not pushed things to the limit to make the program work as fast as possible, the professor of Designed Intelligence explains. The benefit is that the genesis of the composition takes place before your eyes as a real-time process, which presents a beautiful spectacle.
Feijs has been fascinated for a long time by the relation between abstract art and computer science. After having graduated as an electrical engineer he worked at Philips for many years on computer science-related projects before coming back to TU/e at the turn of the century. When he was still employed by Philips, his leisure time was spent for many years on ‘programming’ abstract works of art. In the 1990s he was engrossed in the niceties of what makes a Mondrian a Mondrian. “If you can encompass an abstract work in a computer code, you have learned something about the essence of the work of art. You could regard the code as the DNA of the painting.”
“Of course, you can divide a painting into pixels and store it like a bit map of a couple of megabytes”, Feijs explains, who graduated on computer science theory, “but you can achieve the same result with a handful of bits, which tell you where a line is running and where there is a plane of a certain color.” The real art is to describe the whole painting with as little code as possible. By that I do not mean an exact replica of the Victory Boogie Woogie (that would be too simple and hardly instructive), but a personal composition whose properties come closest to those of a Mondrian. “What you really want to know is how much information there is contained in such a composition. According to Kolmogorov’s information theory the length of the program that you need to describe a work is a measure for the amount of information.”
Many books have been filled with descriptions of the development in Mondrian’s work, and in Feijs’s study we can find a large number of them. “Mondrian went ever further in breaking down the figuration. He painted the same objects time and time again, like trees and churches, and in the process they turned more and more abstract. That is called plastic accentuation.” Nothing was to remind anyone anymore of the ‘real world’. At a certain moment Mondrian only used rectangles and horizontal and vertical lines in his compositions, in the primary colors and black, white and gray. “To rid the composition of all tension, he even stopped using slanting lines.”
One of the challenges of mimicking Mondrian is to create the illusion of a grid, without starting from a predesigned structure. The solution that Feijs came up with for this consists of the random placement of growth nuclei (by means of a random generator), which then develop into lines. “By proceeding like that you achieve the intended grid effect in an organic manner.”
The thickening growth of the canvas with lines and planes is similar to a biological process, like the genesis of bacterial colonies in a Petri dish. Indeed, nature was a source of inspiration for the professor. He even tried via a so-called genetic algorithm to arrive at the most beautiful Victory Boogie Woogie. “To achieve this, I crossed half of the properties of a painting with half of those from another. From the offspring I then selected the best ones, which I subsequently crossed with each other again. Unfortunately, however, evolution unfolds over many generations, and I myself had to act the part of an ecological niche and take care of the natural selection. After a number of nights I’d had enough of that. It was simply taking too much time.”
Still, even without the helping hand of evolution Feijs had managed to produce quite acceptable variants of different Mondrians. He published about his program in such journals as Leonardo, an academic journal for the application of science and technology to visual arts and music. In addition, he made the programming of abstract art part of the ‘Creative Programming’ assignment, which he provides with colleagues for students of his Department of Industrial Design and the Bachelor College. “This project connects our competences of ‘integrating technology’ and ‘social-cultural awareness’. The students are always very enthusiastic about this subject.”
Never did Feijs venture into the Boogie Woogies (except the Victory, Mondrian also made a Broadway Boogie Woogie). Until the Gemeentemuseum in The Hague, which boasts the world’s finest collection of Mondrians –including the Victory Boogie Woogie- held a competition. Feijs saw this as a nice opportunity to get feedback from experts. “Among the members of the jury were three people who had since the 1960s attempted to make this type of generative art themselves, as well as two Mondrian experts who are endeavoring in the Gemeentemuseum to unravel the genesis of the Victory Boogie Woogie by means of X ray and chemical engineering technology.”
So Feijs took up programming again earlier this year. First he dug into the slightly simpler Broadway Boogie Woogie. “When that turned out to succeed quite nicely, I felt confident enough to get started on the Victory.”
The concept of recursion has an essential role in the approach adopted by Feijs. Omnipresent in computer science, recursion is probably known best to the outside world as the ‘Droste effect’, derived from the old-fashioned cocoa tins showing the picture of a lady with a Droste tin on a tray, showing the picture of a lady with a Droste tin on a tray, showing... etc. In computer programs recursion simply means that a function (a piece of code that carries out a certain assignment) invokes itself.
For the Victory Boogie Woogie Feijs programmed the filling of the mini canvases -the planes which in turn are also divided into planes- as if they are paintings of themselves. Or to put it more precisely: each element of the composition (the basic small blocks -atoms-, lines, planes, mini canvases and the whole canvas) is described in a similar way in the program. “They inherit from the same class”, is how Feijs formulates this. He thereby profited from a form of recursion that is naturally present in the composition.
“Unfortunately the computer has no sense of balance and beauty. That is why my program creates the best Victory’s when I carry through the recursion slightly further than in the original. The drawback is that the composition actually acquires too much detail then, although it does enhance the statistic chance that the program generates a composition that is beautifully balanced.”
For the competition, Feijs selected one of the Victory’s that had worked out better and submitted it to the Gemeentemuseum, together with an explanation of his approach. The jury was enthusiastic, the result being that his life-sized entry now stands in Feijs’s study in the Hoofdgebouw. More important, though, is that the Gemeentemuseum had a film made in which elements of the composition are connected with pieces of programming code. This film is shown next to the original Victory Boogie Woogie. “I find that a highly applicable and contemporary approach. And of course it is a great honor as well.”