About generative & digital art (Charles Giulioli, 2019)
Question of style
What makes the “style” of an artist? Is it the expression of an inner self? Is it the fortuitous accumulation of habits and gestures through the years? Would a machine be able to mimic a style? Years before artificial intelligence, I dreamed about such a “Painting Machine”. Would it be possible to feed a computer with enough information to allow it to mimic my own paintings? This question triggered my interest for digital art.
“The Painting Machine”
In 2003, I have submitted my own painting to analysis to create my first computer program, “The Painting Machine”. Being the observer of my own works was very revealing. I discovered that certain elements were determinant while others were secondary. I noticed that over the years I had acquired particular habits and wondered why certain works seemed stronger than others. This led to the description of a process that I could translate into a computer program. The principle is to set precise rules for the overall composition of the picture while leaving many choices at random to the computer. It produces images that look like the works I paint myself.
My surprise was to discover that this part of random in the work is not only a source of fortunate coincidences but also an expression of the universal contingency.
There is something light and joyful in the free appearance of colors and signs. Something which appears purely by chance gives me a feeling of freedom and elation by its mere presence. Similar to what I experienced when I discovered the joyful random of François Morellet’s works or Jean Tinguely’s “free machines “. This irruption of random opens the door to novelty and unforeseen results, an attitude very different from that of a painter.
“Work with no End”
Random has become the driving force for my work. Abandoning a strictly painterly point of view, I attempt to experience the passing of time, the “endless creation of unpredictable novelty” according to Bergson’s expression. In generative works, the image builds up itself in real time. Accidents arise in a random way and the image evolves freely. There is really something going on, there, under our eyes. Movement and rythm play a key role in this work. It is not the recorded image, but its evolution in time that constitutes the work: An attempt to perceive the passage of time as “always the same and always different”.
Little by little, the programs that I write depart from my painting. In “Free Lines”, there is no more traces of my drawings, the shapes are a result of mathematical calculation. On the screen, a line takes shape like calligraphy. My eyes follow the extreme end of this line and its movement tells me something about my life. The mathematical equation gives a “style”, a general coherence to the line whereas the random parameters make it totally unpredictable in detail. This strange compound of structure and improvisation, reminiscent of Jazz music, echoes my own existence.
The capacity of the computer to produce random in a simple way as well as the possibility to create images which evolve in time opens the way to works very different from painting. The quality of light, colors and transparency is also a key feature for my choice. However, it would be naïve to think that the computer creates works by itself. The work is just different from a painter’s. The creation is not in the visible image but upstream to the image, in the program that produces these images. An other characteristic is that the images are produced very quickly and in large quantities. Selection and elimination becomes a key part of the work. I like this part, making a choice between images, wondering why some are appealing to me and others not.
I must admit that computer did not leave me unchanged in the way I see the world around me. Especially I became very attentive to random occurrences.
Material of the Digital
Digital works keep an immaterial character –in spite of the drawings that I use as a base for many programs. One can regret not finding the moving track of the artist’s hand. But one needs to consider that computer sciences have a material of their own. And I believe Jean Dubuffet’s words when he said that “Art has to arise from the material”. The material is unwilling; it questions and challenges the idea of the artist. This is also true of digital art: the material of digital is the computer hardware and the writing of the code. As the sculptor or the painter, the artist experiences limits and resistance. Little by little, he finds a style of his own. He has to embody his intuition in the limited framework of the digital screen and with his own skills to design and develop a code. And the way from the initial idea to the accomplished work is sometimes long and tortuous.
Natural or Artificial intelligence?
Generative works are created by machines, according to rules that include a degree of uncertainty or random so that the output is unpredictable.
In the case of AI, the rules are not established by humans, but elaborated by machines. This is “machine learning”. Through neural networks algorithms, the machine is able to “learn” from thousands of existing portraits what is a portrait and discriminate a portrait from any other image. It is important to note that the rules established by the machine are not expressed and remain inaccessible to humans.
This is all very interesting from a scientific point of view but I find it a bit frustrating for the artist since the exciting part for me is to invent the rules and depart from pre-existing models. And even when my purpose is to mimic my own paintings, as in “The Painting Machine”, it involves a very conscious and deliberate analysis.