Generative Art
This is how I do it
By Stuart Smith
I have the highest admiration for artists who have mastered the arts of drawing and painting, and I’m delighted — and surprised — to find myself among them here on Artique. But I come here with a different kind of art, art that I can create without the manual skills required by traditional art forms. For better or worse, my art falls under the rubric of “generative art.”
What is “generative art?”
Philip Galanter, the leading theorist of generative art, has offered the following definition:
“Generative art refers to any art practice in which the artist cedes control to a system with functional autonomy that contributes to, or results in, a completed work of art. Systems may include natural language instructions, biological or chemical processes, computer programs, machines, self-organizing materials, mathematical operations, and other procedural inventions.” (Generative Art Theory, in A Companion to Digital Art, ed. Christiane Paul, p. 154. John Wiley & Sons, 2016).
This gives a general idea of my practice, but Galanter provides an even more precise characterization of my approach a little later in the same paper:
“In some cases the generative artist creates a system without a pre-existing vision of what the results should be. The artist then explores the system as a new territory and discovers treasures here and there along the way.” (p. 155)
I began developing the app I use to create my art without a clear idea of what the art would look like. Over the last few years I’ve indeed discovered some “treasures” that it can make. I was initially embarrassed about working in this manner, but I was sustained by constant encouragement from my friend and fellow artist, Anna Ursyn. I also felt a lot better about what I was doing after I read Galanter’s paper.
Overview of my Method
My brand of generative art is relatively simple to explain. My app does just four things:
1. It generates source images for subsequent transformation.
2. It performs a sequence of transformations on the source images.
3. It rates each image generated and either keeps it or discards it.
4. It displays the images and, if desired, makes a video of a sequence of images.
Where I cede control to the system is largely in steps 1 and 2. Source images are generated by routines that take one or more random inputs, and each of the transformation routines (a.k.a. “filters”) also takes one or more random inputs. This provides a substantial degree of uncertainty and surprise as to the nature of the resulting images.
Filters
Most of the “interesting” computation is done by the filters. Many of the filters I use would be familiar to users of PhotoShop, GIMP, or Krita. But there are a few others that are relatively uncommon. I use filters in this group frequently. I’ll describe one of the most important ones, called Invert, and then give some examples of its use.
Invert does two things:
1. It reinterprets the x and y coordinates of each pixel in an image as angle and radius in polar coordinates. The origin of the polar coordinate system can be put anywhere you like in the image, but I usually put it either in the center or on one of the corners.
2. It moves pixels near the edge of the image inward toward the center and moves pixels near the center outward toward the edges of the image. It effectively pulls an image inside out.
Figure 2 shows how this works. Figure 2 left is a checkerboard pattern used as a reference image. Notice what happens to the gray and black squares in this image when the pixels are moved.
In Figure 2 center, the pixels from the edges are squashed into small lobes near the center, while pixels from the center are expanded out into much larger areas near the edges. Also notice that the black and white half of the image and the black and gray half of the image both remain in their original locations.
In Figure 2 right, the checkerboard image is deformed counterclockwise around the origin in the upper right corner. The black and gray squares are squashed into a much smaller area, while most of the black and white squares expand into a much larger area.
Some examples
Figures 4–7 are examples of images transformed by the Invert filter. Some additional filters have been applied in each case so that the resulting images are not simply colored copies of Figures 2 center and 2 right. Each image has also been recolored with the palette of a different well-known painting. No content was copied from these paintings; the recoloring operation simply imposes the color scheme of one image on another image.
Figures 4–7 are transformations of source images like the one shown in Figure 3, which is a set of randomly colored horizontal lines.
Figures 4 and 5 were both inverted with the origin in the center of the image. Images of this type often appear to “flow” out of the origin.
Figures 6 and 7 were inverted with the origin in the upper right corner. Images of this type often appear to “flow” out of the corner. Figure 1 was inverted with the origin in the upper left corner. As a result the image appears to flow from there.
©2023 Stuart Smith. All rights reserved.