What is Generative Art?

Art by Gwendal Tsang

We live in the age of the digital image not only because of its powerful hybridity and power to order material reality but also because of its powerful pedesis, interactivity, and power to disorder reality—to create new kinesthetic processes. The beauty of contemporary generative art lies not in its “random” number generators and the sublime affirmation of chaos against the orderliness of contemporary reality. Rather, it is in its capacity to create new kinesthetic processes that play in the complex region between highly ordered and highly disordered images. It gives a high degree of kinetic agency to the matters at work.

The primary question for contemporary generative art is thus how to harness a degree of pedesis in whatever way it can, enter it into an interactive feedback loop, and see where it goes. Humans are just along for the ride. In contemporary generative art, the kinesthetic process itself becomes primary. Subject and object, input and output are folded back over them- selves in an interactive feedback loop to be modulated as a whole, continuous process. This has always been the case in all art to varying degrees, even though most arts have tried to block it and confine it. Today pedesis and interactivity have become a primary and dominant focus of the most cutting-edge aesthetic experiments.

Generative Visual Arts 

In the visual arts, pedetic computer algorithms can be used to produce thousands of iterations with numerous parameters, like color, line length, width, thickness, rotation, texture, distortion, noise, brushstroke, and so on. The artist selects parameters, type of algorithm, and degree or type of pedesis—Perlin noise, loops, iterative variance, and so on. Pedesis can be introduced from the input, process, or output. An incredible variety of stochastic naturalistic processes can be animated, with different results each time. 

In a rapid series of such animations, Maxime Causeret’s Order From Chaos (2016) shows the pedetic patterns of raindrops hitting a surface and spreading, pedetic branching patterns of plants, swarming behaviors of insects, soap-bubble patterns, cellular bifurcations, coral meandering patterns, and more. The images generated are not meant to be copies of natural products but, rather, their own visual expressions of how stochastic algorithms can produce ordered patterns just like nature can, but this time with new resulting organisms. 

More disordered still is Maurizio Bolognini’s Programmed Machines (1988–), composed of enclosed computers generating flows of continuously iterated pedetic images. In the 1990s, Bolognini programmed hundreds of these computers and left them to run ad infinitum. Most of them are still working today. Of these works he says, 

I do not consider myself an artist who creates certain images, and I am not merely a conceptual artist. I am one whose machines have actually traced more lines than anyone else, covering boundless surfaces. I am not interested in the quality of the images produced by my installations but rather in their flow, their limitlessness in space and time, and the possibility of creating parallel universes of information made up of kilometers of images and infinite trajectories. My installations serve to generate out-of-control infinities.12 

In another work, Collective Intelligence (2000), Bolognini used similar machines to project random lines of light onto public surfaces and allowed mobile telephones to interact with them, changing the patterns in real time and creating “generative, interactive and public art.”13 Bolognini thus introduces pedesis and feedback at every level of the aesthetic process. The input is interactive and collective from the population, and the computer processing then randomizes the input, resulting in a highly pedetic and interactive output. 

Radicalizing this idea even further, Scott Draves’s Electric Sheep (1999–) is a computer screensaver that runs iterative fractal flame patterns with a number of different animated parameters. The screensaver is what your computer dreams of while it is asleep, a reference to Philip K. Dick’s novel Do Androids Dream of Electric Sheep? Users can interact with the process by liking or disliking various iterations. This input then reprograms the genetic algorithm as the process mutates to become more interesting to the viewers. Users can also program and upload their own fractal processes whereby they “breed” or mix with the others to produce more iterations. There are currently about 500,000 active users a month.14 Again, pedesis and feedback are incorporated at every level with the aim of finding the most beautiful middle ground of complexity between too much order and too much disorder in the image. 

Generative Literary Arts 

Contemporary generative literary works go beyond the cut-up and fold-in methods of the Dadaists to produce much more pedetic and interactive works than previously possible. Philip M. Parker, originally an affiliate of the Fluxus group, used a mathematical algorithm named “Eve” to produce digital poetry based on graphic theoretical relations between words in the dictionary. He has produced over 1.3 million poems in this manner. He has even used similar algorithms to produce entire books—200,000 of them.15 

More recently, Jason Nelson has used generative methods to create digital and interactive hyperpoetry. His famous “Game, Game, Game And Again Game” (2007) uses flash media to create an audiovisual mashup of text fragments, sounds, and video in an interactive video game format. “I made this. You play this. We are Enemies” (2009) develops the same idea. His “Uncontrollable Semantics” (2006) creates a series of words on the four corners of the screen, each with its own sound and image. As one clicks on the different words, new word–image combinations are created. Poetry becomes a series of continually modulated feedback loops. A similar feed- back loop of interactive options occurs in Neil Hennessy’s “JABBER: The Jabberwocky Engine” (2000), in which randomly floating letters are connected to form new combinations of neologisms that produce pro- nounceable English words, but with no dictionary definition. These are then incorporated into poetic works. 

Jean-Pierre Balpe has even produced stochastic and interactive novels such as Trajectoires (2000) and Fictions d’Issy (2005) by using algorithmic and interactive methods. The stories are continuously generated sentence by sentence, and readers can shape the outcome by using their phone’s keypad. Balpe’s work and many others are contained in the first volume of the Electronic Literature Collection (2006) and they represent an amazing diversity of generative literary works.16 All these give the materiality of words a maximum of pedesis and interactivity by tying them to the kinetics of the digital process. 

Generative Plastic Arts 

With the advent of 3D printing, generative algorithms can now be modeled directly into plastic media. Although the technology is still in its in- fancy, some of the initial creations are incredible. Among the most amazing examples are the sculptures and architectural columns made by Michael Hansmeyer (figure 16.4). Hansmeyer is an architect and programmer who uses algorithms and computation to generate unique architectural forms using a simple feedback algorithm of topological folding. Hansmeyer begins his designs with a single cube and then begins to stretch and bend the cube, applying his folding algorithm to different parameters such as depth, curve, and line. The results are incredible—forms so complex that the “artist” could not possibly have “an idea” of them. The whole matter-form distinction collapses onto itself as matter becomes morphogenetic and semi- autonomous. According to Hansmeyer, 99 percent of the algorithms end up producing noise. Only those with certain modulated parameters produce the most complex forms. In addition to the Doric, ionic, Corinthian, and undulating orders of columns, Hansmeyer has produced an entirely new architectural order: the generative order.  

 

Figure 16.4 Michael Hansmeyer, Columns (2010)
Source: From artist’s website, © Michael Hansmeyer, http://www.michael-hansmeyer.com/projects/columns. html?screenSize=1&color=1#1. 

Nervous System, a generative design studio, uses algorithmic and stochastic code to create unique sculpture, jewelry, light fixtures, and even clothing using 3D printing. Their Floraform sculptures are similar to the biomechanics of growing leaves and blooming flowers. Their Xylem (2D) and Hyphae (3D) sculptures use algorithms that produce structures similar to those found in the veins of leaves. These patterns are used to generate jewelry, lamps, sculpture, and even architecture. Their Kinematics sculptures add a fourth dimension to 3D printing by creating a design system of hinged panels with a simulation strategy of folding and compression to produce customized designs that can be fabricated efficiently by 3D printing. The structure is printed as one part, but has thousands of interconnected pieces that require no assembly. The result is kinetic dresses, lampshades, jewelry, and more. 

Additionally, their website includes interactive software that allows an- yone to design his or her own sculptures and print them. Nervous System’s designs thus use hybridity to physically transcode binary code into 3D and 4D sculptures. They use pedesis in their stochastic algorithms, and they use interactivity in the user interface which is sensitive to its initial and continuous conditions. The purpose is not simply to maximize noise or feedback or to copy natural patterns but also to produce new patterns through the modulated use of noise and feedback. The purpose it to give the electromagnetic field its maximal kinetic and material generative agency. 

Generative Sonic Arts 

It was Brian Eno in the 1970s who first coined the term “generative music,” but the scene has expanded dramatically since then. Today, generative music has vastly outstripped Mozart’s dice throws, futurist noise music, and even the later modernist aleatory music of Cage, Feldman, Boulez, and others. These earlier works relied on comparatively simple pedetic parameters and limited feedback systems, and they remain but modest precursors to the much more hybrid, pedetic, and interactive works of generative music today. 

Some of the first works to introduce a higher degree of pedesis and feedback were Stockhausen’s Kontakte (1958–1960), Terry Riley’s The Gift (1963), Brian Eno and Robert Fripp’s No Pussyfooting (1973), and Eno’s Discreet Music (1975), the latter which used a new tape-loop feedback system combined with an echo unit and a continuously modulated graphic equalizer to change the timbre of the sounds. This allowed sound to turn back over itself in an ever-expanding and interactive modulated feedback pattern of sonic images. Similar modulated tape-loop feedback systems continue to be used today by various ambient music artists, such as Ous Mal, Taylor Dupree, Tape Loop Orchestra, and William Basinski. For con- temporary musicians, the tape-loop process also introduces a new focus on the pedetic sound of the tape noise itself. 

Although present in Eno and Riley’s early work, and emphasized in works like Steve Reich’s amazing Pendulum Music (1968), which swings microphones over speakers generating patterned yet chaotic feedback, contemporary artists have turned increasingly toward the stochastic noise, feedback echo, and hiss of the tape itself—amplifying it, looping it, and dramatizing the noise of the electromagnetic field. This is part of a much wider trend by contemporary generative musicians to seek out pedetic sounds like tape hiss, noise, vinyl-record crackle, CD skipping sounds, microphone feedback, FM radio static, and other irregular, pedetic and traditionally undesirable musical sounds created by the pedesis of the EM field. The aim is not simply to reproduce these sound images but also to work with them and use their stochastic patterns as the basis of new feedback loops and patterns of their own. 

In The Caretaker’s An Empty Bliss Beyond this World (2011), for ex- ample, vinyl crackle is amplified and echoed to the point where it equals the volume of the looped vinyl melodies. In a slightly different vein, Burial’s Burial (2006) uses the static crackles and pops reminiscent of those that occur in maxed-out speakers and loose or old audio cables, or the static electricity pops from the audio mixing equipment and microphone itself. These crackles become the sonic milieu of his hyperdub loops. Glitch albums like Oval’s OvalDNA (2011) combine various melodic audio feedback tones with CD skipping noises, as if one had taken a knife to a CD’s surface and then stuck the CD back in the player. The use of FM static in Olli Aarni’s Pohjoisen Kesä (2012), or his use of field recordings of underwater insects in Vesiä (2017), or Mileece’s interactive bioelectrical feedback sounds gathered from plants all accomplish the similar aim of introducing pedesis into the audio feedback loops for sonic modulation—to give noise “a life of its own,” to paraphrase Pollock.

Even more dramatic, however, is the use of numerous types of digital pedals, oscillators, tone generators, and computer software to produce highly diverse and numerous loops of sound that can all be modulated in medias res and with more technical precision than any tape-loop audio noise. The famous Japanese noise musician Masami Akita “Merzbow” has produced particularly pedetic and abrasive albums such as Pulse Vegan (2014), using both granular synthesis software and numerous digital sound boxes or pedals. In his most recent work, the software transforms his sounds into “clouds” or flows of micro sounds that can then be modulated continuously and generatively as a whole, according to a number of different parameters and computer algorithms. 

Curtis Roads, a media arts professor and composer of Point Line Cloud (2005), describes the process in fluid dynamic terms: 

Beneath the level of the note lies the realm of sound particles. Each particle is a pinpoint of sound. Recent advances let us probe and manipulate this micro acous- tical world. Sound particles dissolve the rigid bricks of musical composition— the notes and their intervals—into more fluid and supple materials. The sensations of point, pulse (series of points), line (tone), and surface (texture) emerge as the density of particles increases. Sparse emissions produce rhythmic figures. By lining up the particles in rapid succession, one can induce an illusion of tone continuity or pitch. As the particles meander, they flow into liquid-like streams and rivulets. Dense agglomerations of particles form clouds of sound whose shapes evolve over time.17 

Granular or pulse software thus introduces into music a new fluid dynamics of flows to the sonic image, letting it pedetically meander into periodic densities or folds that are then woven into a larger sonic texture like a fabric. However, the term “grains” of sound is misleading because each micro 1- 50ms sound sample or “grain” is buffered by an amplitude modulation or “envelope” that connects the grains in a sonic continuum. Wave-scanning techniques can also eliminate the need for the envelopes by having the grain boundaries always meet at the zero-crossing point of the respective signals. The resulting composition is thus sonically continuous and has a highly fluid character to it like the sound of rushing water, crashing waves, or a turbulent dripping faucet. Barry Truax’s Riverrun (1986), for example, is a direct statement on the fluid dynamic nature of micro-sonic generative image composition. 

“From the smallest rivulet to the fullest force of its mass, a river is formed from a collection of countless droplets and sources. So, too, with the sound in this composition which bases itself on the smallest possible ‘unit’ of sound in order to create larger textures and masses. The title is the first word in James Joyce’s Finnegan’s Wake.”18

Such modulation was impossible with the instruments and techniques available before the late twentieth century. For the first time ever, it is possible to modulate noise-pitch-rhythm as the complete sonic continuum that it is, at the smallest possible audible levels of the waveform, thus introducing an incredible new range of pedesis. Recent works integrating granular synthesis also include Ian William Craig’s Centres (2016), Kaitlyn Aurelia Smith’s Ears (2016), and Multicast Dynamics’s Scandinavia (2016). Mixing various higher degrees of pedesis into the digital input, pro- cess, and output produces numerous genres and subgenres of electronic, electroacoustic, and experimental music: glitch, drone, ambient, postclassical, noise, tape music, field recordings, found music, circuit bending, sound sculpture, vaporwave, chopped and screwed hip hop, and many more. 

Additionally, contemporary generative music introduces a new level of hybridity and feedback never before possible in music. Brian Eno’s latest album Reflection (2017), for example, is a brilliant mixture of pedesis, hybridity, and feedback. The album uses stochastic algorithms to determine the parameters of the sounds.

Because everything in the pieces is probabilistic and because the probabilities pile up it can take a very long time to get an idea of all the variations that might occur in the piece. One rule might say “raise 1 out of every 100 notes by 5 semitones” and another might say “raise one out of every 50 notes by 7 semitones.” If those two instructions are operating on the same data stream, sometimes—very rarely—they will both operate on the same note . . . so some- thing like 1 in every 5000 notes will be raised by 12 semitones. You won’t know which of those 5000 notes it’s going to be. Since there are a lot of these types of operations going on together, on different but parallel data streams, the end result is a complex and unpredictable web.19

Second, the album uses an interactive process of modulation as Eno “tweaks” the parameters during playback over and over again.

Pieces like this have another name: they’re GENERATIVE. By that I mean they make themselves. My job as a composer is to set in place a group of sounds and phrases, and then some rules which decide what happens to them. I then set the whole system playing and see what it does, adjusting the sounds and the phrases and the rules until I get something I’m happy with. Because those rules are probabilistic (—often taking the form “perform operation x, y percent of the time”) the piece unfolds differently every time it is activated. What you have here is a recording of one of those unfoldings.

Third, the album uses a hybrid transcoding of the music into an audio-visual- haptic software application that allows users to touch a colored screen and modulate the endlessly looped stochastic patterns for themselves.

REFLECTION is the most recent of my Ambient experiments and represents the most sophisticated of them so far. My original intention with Ambient music was to make endless music, music that would be there as long as you wanted it to be. I wanted also that this music would unfold differently all the time— “like sitting by a river”: it’s always the same river, but it’s always changing. But recordings—whether vinyl, cassette or CD—are limited in length, and replay identically each time you listen to them. So in the past I was limited to making the systems which make the music, but then recording 30 minutes or an hour and releasing that. REFLECTION in its album form—on vinyl or CD—is like this. But the app by which REFLECTION is produced is not restricted: it creates an endless and endlessly changing version of the piece of music.21 

Reflection is thus an attempt at mimesis of neither natural products nor natural processes but, rather, a way of becoming what it is: matter in motion—pedetic, hybrid, and interactive. Just as the flow of matter has no beginning and no end, neither does Reflection. The three creative stages Eno describes for this work match up directly with the kinetic ones laid out in this book: (1) pedetic material flows intersect at a constellation, (2) fold into a distribution of affective loops, and (3) are continuously modulated as a whole woven field of sound. Eno writes,

The creation of a piece of music like this falls into three stages: the first is the selection of sonic materials and a musical mode—a constellation of musical relationships. These are then patterned and explored by a system of algorithms which vary and permutate the initial elements I feed into them, resulting in a constantly morphing stream (or river) of music. The third stage is listening. Once I have the system up and running I spend a long time—many days and weeks in fact—seeing what it does and fine-tuning the materials and sets of rules that run the algorithms. It’s a lot like gardening: you plant the seeds and then you keep tending to them until you get a garden you like.22

Numerous other efforts to increase the interactivity and hybridity of music abound. Media artist Scott Snibbe, for example, has created a number of such interactive music album applications, like Bjork’s Biophilia (2011) and Metric’s Synthetica (2013). Snibbe’s app Motionphone (2012) integrates sound, kinetic motion, and visual animation. As users move their fingers across the screen, their movement is animated and looped. These can then be shared and interact with other users’ kinetic sculptures online.

Read more from Theory of the Image

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s