Generative art wasn’t born yesterday. Artists have used structured, systematic methods for making art since at least the early 20th century, an interest that was revived in the postwar period and quite literally electrified as artists gained access to electronic computers in the 1960s. Six decades later, generative art is back. In this virtual roundtable, I talk to a new wave of “generartists” — six leading creators in the fxhash space.
My questions are shaped by my own art historical research on Vera Molnar and early generative art. While it’s clear that generative art is light years ahead of where it was, in terms of its technological advancement and its accessibility to both makers and viewers, it’s striking to see how artists are exploring many of the same themes and questions as Molnar and her peers. Read on to hear Robert Hodgin, Ciphrd, Aleksandra Jovanić, Lunarean, Lisa Orth, and Ivona Tau discuss what generative art means to them, how it relates to abstraction, and where they see it moving in the future.
Zsofi Valyi-Nagy: How do you define generative art, and what does it mean for your practice?
Lunarean: To me, generative art is when you create a system that itself can produce art. For example, it could be an algorithm implemented as computer code. It might include an element of randomness so that it results in a different output every time it runs, which means a single algorithm could in theory produce an infinite number of artworks. I enjoy it because it’s capable of creating outputs I would have never dreamed of and when I work on a generative algorithm, I sometimes feel like I’m just exploring an infinite space of art.
Lisa Orth: I define generative art as art that’s created using a set of rules — be they algorithmic systems or mechanical processes — and incorporates random functions, elements, or autonomous processes, to determine artistic outputs. This could include something as simple as rolling dice to determine color usage, or as complex as creating entire algorithmic systems that interact with one another. For my practice, I love the idea of using the random() function as a way of collaborating with the universe — the element of chance used as a channel of communication to something outside of yourself and your own ego.
Robert Hodgin: I think of generative art as taking a half step back from the canvas. More traditional art forms (painting, sculpture, photography, etc.) are often about embracing spontaneity and recording the “now” either through brushwork, carving, or shutter release. The generative aspect of my art is about the moments before the “now.” It is about designing and refining systems before releasing them and observing how they make their own marks on the canvas. It is a collaboration between the spontaneous and the calculating sides of my artistic expression.
Alexandra Jovanić: I always prefer inclusive, rather than restrictive definitions. I love to reinterpret Galanter’s definition in free form that generative art is where the artist creates an autonomous system that when put in motion produces (audiovisual) output.¹ This means that I’m developing algorithms with a set of constraints, and the computer will execute code and produce an infinite number of outputs. Sometimes it provokes emotion or communicates the artist’s idea. If constraints are narrow, outputs will be similar. Making an algorithm and finding the right boundaries is my practice.
Ivona Tau: In a very broad sense, generative art is any kind of art that was created by following a procedure or algorithm. While, in the past, forms of generative art were traditionally defined by very precise instructions, the evolution of AI allows for a less controlled generative process, guided by optimization algorithms instead of strict rules. For me, this brings a vast unknown to the creative process. While I’m able to control the general shape of the distribution of generated images, in no way am I able to define the boundaries and limits of what gets created.
Ciphrd: To me, generative art is a way to express my algorithmic ideas without the usual constraints of software development in the context of “utility.” It’s a way to explore freely the algorithmic space knowing that I only need to deliver for myself. As for a more generic definition, Wikipedia’s aligns with my own: “Generative art refers to art that in whole or in part has been created with the use of an autonomous system.”
ZSVN: Which artists or thinkers have influenced your thinking about generativity?
LO: Vera Molnar, Marcel Duchamp, Brian Eno, William S. Burroughs, Hilma af Klint, the Surrealists, Spiritualists, and other movements utilizing automatism in their work. Of the artist’s use of intuition Molnar has said, “There is this thing that can replace intuition. It’s randomness.” That statement is so powerful for me within the practice of generative art. I hope not to replace intuition with generativity, but instead to build upon it.
AJ: I often return to Sol Lewitt’s work. His instructions for wall drawings were simple and straightforward, written on paper. With Lewitt, you were collecting this piece of paper rather than the actual rendering. His “code” could be executed by anyone — allowing the piece to adapt to different hands and situations. He also opted for drawing directly on the wall in order to remove the medium. He used combinatorial systems to create complex artworks from simple components. Behind every generative art piece is a simple algorithm.
L: Early on, I was most influenced by Jared Tarbell’s work, which often uses simple rules that somehow lead to intricate results. This has led me to often start a project by playing with an algorithm that’s simple but highly generalizable to see what emerges, rather than trying to build something with a particular aim in mind.
C: It’s hard for me to reference particular people, as my thinking has emerged gradually over the last decade, inspired by many actors from a wide variety of fields. Artificial life is a field close to my heart and probably my main source of algorithmic inspiration, whether it’s Tierra or Clusters. I’m also addicted to microscopy and the tools developed for that purpose. I love observing microorganisms in their highly mechanical yet organic behaviors. Needless to say, this is my main source of visual inspiration.
I’ve also been highly influenced by video games — their various systems, the optimization of every line of code, the architecture of an application, the rendering quality, the user interactions. Beyond that, software design opens up the mind to many design paradigms on how to generate and process data. It helps to make an idea reality.
RH: Early on, when I was first exploring how to create with code, my inspiration came from people who were incredibly skilled with the medium. Casey Reas, Jared Tarbell, and Toxi were all very important to my growth as a digital artist. As I became more comfortable with the medium, my inspiration started to come from sources outside the world of computer coding — the sciences, natural processes, as well as traditional painting, watercolor, sculpture, and photography.
IT: I’m deeply moved by the conceptual artists who were working with procedures in the age before computing. Learning about Sol Lewitt’s work shifted my perception about generative art to a more human-centric interpretation, where art is defined by an intangible idea rather than the final execution. I’m also inspired by Vera Molnar, Manfred Mohr, Roman Verostko, and the “Algorists,” though my biggest influences actually come from experimental and Surrealist photography. The shift toward generating one’s own reality with a camera, instead of mimicking the surrounding world, built solid foundations for generative art. Indeed, generativity is also the ability to abstract meaning from ordinary objects, as was done ingeniously by Man Ray and Dora Maar.
ZSVN: Much of generative art could be described as abstract, in the aesthetic sense of nonfigurative. More than on any other NFT marketplace I’ve seen, scrolling on fxhash, one finds a lot of patterns, textures, and shapes that don’t necessarily represent anything; and even if they do — a plant or a mountain, for instance — the referent tends to be quite abstracted. Would you say you are drawn to abstraction? If so, what role does it play in your approach and/or in generative art more broadly?
AJ: That is interesting because I see my pieces most of the time as borderline figurative. When abstract shapes remind me of something, I take them to the next level and redesign them a bit. Luckily, there is always a limit to how much you can do with pure code. Simplified illustrations, smooth shapes, and line drawings also play a part in my works in other media.
RH: The abstraction is born of necessity. We are creating with lines of code, and the easiest things to make with code are going to be basic shapes: Points, lines, and triangles. The works we see on markets like fxhash are generally highly iterated explorations of points, lines, and triangles that evoke emotion and significance. In NFT markets that lack a code-based requirement, you will see much more representational art created with 3D programs. In places like fxhash and Art Blocks, the code is the palette from which we create.
C: In its essence, generative art comes together with some level of abstraction, especially when using algorithms. In order to draw on a digital canvas with code, it’s necessary to use mathematical abstractions in the first place, whether rectangles, lines, or functions of any kind. As a result, we often recognize geometric primitives in works of generative art. Indeed, it’s very difficult to generate realistic imagery purely from code, typically requiring many years of practice. Most of the time, it’s easier, and just as interesting, to explore a more abstract aesthetic.
Each algorithm introduces its own universe, made of its own rules, and it definitely makes sense to explore it on a purely abstract level, both conceptually and visually. I try to find small and elegant rules inspired by nature and apply them to many individual entities of a virtual ecosystem. In the process, some kind of life emerges. I’ve tried to refine my aesthetic to give it some level of realism, for viewers to connect with the artificial life they observe.
L: A lot of the algorithms I use and find most interesting are ones that lead naturally to abstract outputs. So in my past work, when the outputs do represent something, they do so in a minimal way, such as representing buildings with straight lines. I avoid adding more detail than necessary to keep the algorithm as simple and pure as possible. I find that the more you try to make the outputs look like something, the more you need to complicate the logic and restrict the space of possible outputs.
IT: I’m drawn to the generalization of ambiguous shapes. A big part of my process is taking pictures of the surrounding world and feeding them into neural networks, in order to explore how the meaning and individuality of the images is lost. It is a way of transforming my personal memories into a generalized vision that viewers can relate to in unexpected ways. With Study of atom in analog, digital and A.I. (2022) I take this process to the extreme by training AI on the fragments of old photographs, which produces abstract and ill-defined training data. The goal is to find the building blocks of reality, based on pieces that are illegible individually but when taken together highly personal.
LO: Abstraction facilitates creative journeys of pure exploration to destinations I most likely would never have arrived at through figurative imagery. Abstract art is a way to delve into areas that my conscious mind isn’t easily able to explore. I love blind exploration, journeying without a preconceived objective.
When a color or shape moves you deeply, it’s like being privy to an act of pure creation — like watching the birth of a planet or a new life form. But I realize the irony in using emotional concepts to talk about art created with non-emotional concepts or processes. As my approach to creating generative art tends to be more exploratory rather than goal-oriented, abstraction is stylistically an ideal fit.
ZSVN: Is a generative algorithm more of a tool or a collaborator in your practice? Could you speak about the relationship between your input as an artist and the autonomy of the technology? How much do you intervene in the results, and on what basis? And lastly, in the realm of digital and generative art, how important is it for an artist to write their own code?
C: Algorithms can take the form of both a tool and a collaborator. At first, there’s the question of what might emerge from a particular system. Then comes its implementation and exploration. We always discover surprising results along the way simply because our brain cannot predict all the parametric spaces that a system might produce. Whether it’s bugs or unexpected behaviors, there’s always something that drives our original intent towards a new idea. Designing a generative algorithm is a back and forth between the algorithm, the output, and an eventual selection based on our appreciation of the outcomes. I think as we learn to tame this process, we necessarily fall in love with it because it’s a perfect mixture of control and surprise.
I think writing your own code is not a requirement, but in some instances it’s necessary to remove a barrier. I like my pieces to evolve in real time so that viewers can interact with the simulated ecosystem as if they were studying species in real life. To achieve that, it’s important to maintain close control over each instruction being executed by the machine, from the CPU [Central Processing Unit] to the GPU [Graphics Processing Unit]. I must code some components myself to achieve that.
AJ: A generative algorithm is more of a tool, and a little less of a collaborator. More often than that I’m willing to admit, I’m pleasantly surprised with the output from my algorithms. I see it as magical, but eventually, like a magician’s trick, I understand how the computer got that visual result. I do intervene with the results, either by cherry-picking outputs or by fixing algorithms to produce results I will eventually like. It is crucial to write your own code. During the process, new ideas will often emerge and your own bugs or missing boundaries take you to unexplored areas.
RH: It’s a collaborator. I spend as much time refining the tools as I do critiquing the output. It is a constant back and forth. Some days, I’m neck deep in the code. This is when I am designing the inner workings of the algorithm. I then switch over to the more artistic side of this collaboration.
I think it is incredibly important for artists working in this specific space to write their own code, or at least to collaborate (and acknowledge) whoever wrote it. One of the difficulties of marketplaces like fxhash is the vigilance required to keep people from exploiting others’ hard work. When you create and upload code-based projects, it is easy to see the code if you wish to. This means that it is also easy to take this code, tweak a couple of variables, and re-release the code as a new project. However, this is also a gray area. I create with p5.js, which is a whole framework filled with code I did not write. Ultimately, it comes down to intent and it can be very difficult to assess intent.
L: It can actually be both. At the start of a project, I’d say it’s more of a collaborator. The algorithm can produce results I don’t expect and those can end up as a source of inspiration. Once I have a more concrete idea of what I’m trying to create, I tend to use code more as a tool. For example, if I want some lines to look pencil-drawn. It’s definitely useful to be able to write your own code, as it provides the most flexibility and control over your algorithm.
LO: I view generative algorithms more as tools than active collaborators. I look at code as another artistic tool. Where can I get to by using it? What can I uncover in the process? I like to allow the code as much space as possible before I intervene in a curatorial way, and I often explore pathways not initially taken.
I suppose you could get by as a digital artist and not write your own code, but for generative work I really don’t think that’s the case. Granted, I’m using p5, so I need to write the code myself, there’s no graphical interface allowing me to drag and drop elements or anything like that. I find that when you have an understanding of how your tools work, and have the ability to create your own tools, that your art will be a truer reflection of your own unique inner voice. That knowledge helps to elevate our craft and gain a better understanding of the mediums we’re working with.
IT: For me, generative art is a form of modern watercolor. When the result of a procedure exceeds my expectations, by creating something that seems out of the distribution and outside of the rules that I have defined, I am amazed to witness unexpected creativity being born. That said, when using random distributions or working with deep neural networks, I do not have complete control over generative outputs, just as I have no power over flowing water in watercolor.
ZSVN: fxhash seems to be invested in the longevity of generative art as well as the concerns of artists. I’m curious to hear your thoughts on smart contracts and how you learned about them. There seems to be a learning curve involved in writing them, so I wonder if you could offer some advice to artists out there who want to future-proof their NFTs. Also on the question of longevity: If a major art museum were to buy your work, what would you want them to know? And how would you want your work to be shown?
RH: I learned about smart contracts like most people probably did, on Twitter. I didn’t end up doing much in this new world of NFTs until I spoke with a couple of friends that were beginning to put work up on fxhash. They convinced me that it wasn’t as scary as I made it out to be and that I would probably enjoy the experience of creating generative art for the blockchain.
On longevity, I have adopted a stance influenced by my early experiences with creating digital art: It can easily disappear. All of my personal Flash portfolio disappeared in a Zip drive failure two decades ago. Because of this, I have a more pessimistic view of digital art’s lifespan.
AJ: I still consider myself a noob in this field. I’m still learning and exploring, so I don’t feel comfortable sharing advice or opinions on smart contracts and future-proofing NFTs.
For years I’ve been working as a web designer and teaching web design and development, so I’ve got used to the fact that you can’t control everything. Owing to the many different platforms, apps, and display sizes, one can only assume how the user will experience your design or artwork and hope for the best. In some of my previous larger projects, I put a focus on experiencing art as a private act in a private, comfortable space. If interaction with a work is necessary, people tend to interact more freely when they’re not being watched. But I don’t have a universal answer on how my work has to be shown; every piece is different, and there is always more than one option.
L: Most platforms allow you to upload the source code for generative art without having to write the smart contract yourself. However I think it’s still important to be aware of how each platform stores and executes your code. For example, some platforms upload the source code to IPFS, with only a link stored on the blockchain. Others store both the source code and the randomness seed on the blockchain, which has the benefit that artwork can be generated solely from on-chain data. However, a more robust way is to store it “in-chain,” where the smart contract contains the actual instructions to generate the artwork, meaning no other tool or language is required to produce the artwork.
C: If I have one piece of advice for artist-developers with any level of experience, it is to overcome intimidation. Because smart contracts manipulate money, it’s quite scary to think about deploying your own. However, I would say it’s even simpler than writing most other types of programs. That said, it’s important to learn the basic concepts properly — how are smart contracts executed? How are they stored? How are they storing and reading data on the blockchain? What are the different data types? They must be designed carefully to avoid any unexpected behavior.
If a major art museum were to buy my work, first of all it would be cool. I would like them to know the technical requirements to display the piece in the best possible manner. I would also love the visitors to experience the kind of interactivity intended in the original piece. They should also have access to a basic explanation of the underlying algorithmic systems behind the work — not an in-depth technical description, but rather an overall conceptual one.
LO: I’m learning about smart contracts a little at a time. At first I didn’t even understand what metadata was, or why it’s important. Can I be an artist creating NFTs and not understand the technology? Sure, but how much more can I do if I understand how it works and do it by myself. When artists are able to control every aspect of their art, from creation to distribution to display, we come closer to communicating our artistic vision in its purest form.
I personally would love to be able to code my own smart contracts to mint NFTs on my own site and build my own marketplace, with the ability to integrate those with existing collections and marketplaces. Do I need to learn how to do this? Of course not. There are so many options already available for me to use. Perhaps I don’t want to depend on someone else, or feel like someone else has more control over my art than I do. I remember the day that I woke up and the Hic et Nunc site just wasn’t there anymore; it really hit home that my art was in someone else’s hands. It was a good wake-up call. I think artists could start to future-proof their NFTs by pinning their metadata to IPFS. That would be a great start, and it would do a lot to validate collector trust.
If a major museum were to purchase my work I hope they would not only have an understanding of the tools used to create the art, but also the broader history of generative and abstract art, as well as NFTs. My pieces aren’t interactive or animated, so I’m imagining their display would be relatively straightforward — high-resolution display in the work’s intended aspect ratio.
IT: I got introduced to smart contracts as a by-product of following the works of earlier AI artists such as Mario Klingemann and Anna Ridler. I am far from an expert on writing smart contracts, but I try to think of ways that a smart contract could bring an additional dimension to my art.
A major revolution would be for museums to depart from established norms on how art should be displayed. I see far too many attempts to show digital art in the same dimensions as paintings or photographs. If a major art museum were to exhibit my work, I would like my art to be shown on a bigger scale, in an immersive setting. A lot of my works are about emotions perceived by viewers and, for them to have the full effect, it is essential to be as close to the art as possible.
Robert Hodgin is a digital artist living in Brooklyn with his husband and two cats. He is a co-founder and partner of Rare Volume, a design and technology studio with offices in New York and Austin. Hodgin’s work is featured in the V&A, Smithsonian Design Museum, The Exploratorium, Wired, and Wing Luke Museum. He is currently working on a series of physical prints and exploring new concepts for upcoming releases.
Aleksandra Jovanić is an artist and programmer from Belgrade, Serbia, who holds a PhD in Digital Arts and a BSc in Computer Sciences. In her research and artistic practice she combines various media, mainly focusing on interactive art, art games, and generative art. As an assistant professor, she teaches at the new media department at the Faculty of Fine Arts in Belgrade.
Lunarean is a developer based in London, UK who enjoys creating generative art in his free time.
Ivona Tau is a generative AI artist from Vilnius, Lithuania who works with neural networks as a medium in experimental photography and motion painting. She is currently pursuing a PhD in Computer Science in the Multimedia Department of the Polish-Japanese Academy of Information Technology. Her work has been auctioned at Sotheby’s, Durán Arte y Subastas, Beijing Poly, and showcased worldwide. Tau is currently a TOP 10 female creator on the Tezos blockchain.
Zsofi Valyi-Nagy is a Hungarian-American artist and art historian currently based in Berlin, where she is a DAAD fellow at the Institute for Media Studies at Humboldt University. Here, she is using media archaeological methods to develop her PhD dissertation on Vera Molnar and the intersection of abstract art and early computer graphics in Europe. She is also a predoctoral fellow with the National Gallery of Art in Washington, DC and a PhD candidate at the University of Chicago. She wrote about “crypto lingo” for the first edition of Right Click Save and is currently writing a dossier for HOLO about Molnar’s exhibition, “Variations,” at the Beall Center for Art + Technology.
Danielle King is the CFO/COO of ClubNFT, and the former manager of the Department of Painting and Sculpture at the Museum of Modern Art.
¹ “Generative art refers to any art practice where the artist uses a system, such as a set of natural language rules, a computer program, a machine, or other procedural invention, which is set into motion with some degree of autonomy contributing to or resulting in a completed work of art.” P Galanter, ‘What is generative art? Complexity theory as a context for art theory,’ 6th Generative Art Conference, January 2003.