An Interview with Ganbrood

Alex Estorick: Could you introduce yourself as an artist, and as a human who collaborates with machines?‍

Ganbrood: I was born and raised in Amsterdam and I have always been engaged in activities that circled around art, but I have only been working with autonomous systems for a short time. I feel like a fast-growing, sci-fi test-tube baby. I have always been interested in visual storytelling, illusion, and special effects, whether for film, commercials, or video games. Throughout my career I have always been trying to synthesize life.

‍AE: Tell me about your work before the NFT came along and before GANs (generative adversarial networks) entered your life.

‍G: I started out as a CGI artist involved in special effects compositing and 3D animation until I became an art director for Sony Interactive Entertainment. I then left the digital realm and worked for 14 years as a documentary and reportage photographer, using no special effects or digital enhancements. Three and a half years ago, a friend of mine showed me Ganbreeder (now Artbreeder) which was a very easy interface for playing around with AI, neural networks, and GANs. I was immediately hooked. 

From the first day, I played around with algorithms that produce synthetic images which seem to live on the edge between illusion and reality. That is where I always like to be and everything I’d ever done up to that point seemed to point toward this.

There was also an element of serendipity because the pandemic had left me without contract work as a photographer. It also coincided with the emergence of NFTs. At the time, I tried to get on some Ethereum platforms, but I read about it being bad for the environment and decided it might not be for me. Then some artists that I was following on social media — Mario Klingemann, Memo Akten, and Joanie Lemercier — suddenly started talking about Tezos as a great alternative. So I joined Hic et Nunc in its second week and I’ve been rabidly engaged with the community ever since, both as an artist and as a collector.

‍AE: Early GAN art always seems derivative of human creativity, but your work conjures worlds that are fundamentally hybrid and chimeric, implying nonhuman creativity. 

‍G: I’ve actually always been intrigued by pure styles, whether Greco-Roman, Jugenstil, or comic art. Somehow I find in GANs — we call it now diffusion — a way to tie everything together in a coherent way that aligns with my personal fantasies, tastes, and ideas. 

The whole thing is an exploration of my inner soul where it touches the outer world, cultures, and everything I’ve learned and seen. 

The beauty of that process is that it’s very similar to what is happening in a deep learning model — the idea that you feed a machine millions of images and see what it does with them.

‍AE: Can you describe the technical approach you adopt? ‍

‍G: I’m not a coder. As a photographer, I had colleagues who were keen to embrace the complex settings of digital cameras, whereas I relied on my intuition and a couple of buttons. I’m not interested in what happens under the hood. I’m very interested in what I can do with the technology. With GANs, I found that the best way to make them do what I wanted was not by rationalizing them but by relying on my intuition.

There was a point when I was looking into training my own model, considering whether to use royalty-free images or my own photography. But it felt like a race against the clock, because the technology was exploding and the results produced by those who were doing their own training were interesting but always below par compared to what I got from what was available. I have always used open source and freely available models. And by combining and layering them I got the results I wanted. 

‍AE: In principle, writing code is also a means of preserving a level of human control over generative systems. What kind of agency do you retain when working with tools like Stable Diffusion?

‍G: When we reached the point when anyone could utter a few words and produce something, that frightened me because I wondered if people would start doing what I do, but better. 

It took me a while to realize that most of what is appreciated in my work is me, not the machine. That imposter syndrome is a large part of being an artist. 

I didn’t use DALL-E for a long time because they kicked me off right away. I also felt very obstructed by working with an algorithm that has taboo words and censorship. Stable Diffusion was a great way to escape that. 

‍AE: There’s an emotive quality to your work that defies its machinic production. Works like Ill met by moonlight, proud Titania (2022) recall landscapes of the Dutch Golden Age, while conveying a mood similar to Arnold Böcklin. Its title is also based on Shakespeare’s A Midsummer Night’s Dream. Can you speak about your historical reference points and the romanticism that seems to pervade your practice?

‍G: I think that the artist’s role is to communicate their entire soul to the audience. I’m using these tools to show people what’s inside of me. The work in question started with a photograph I took in the northern part of the Netherlands, in a place that has a deep meaning for me. I then combined it with colors, shapes, and a story that resides somewhere in my soul. I am a romantic and I love beauty, but at the same time, I know that making a nice picture isn’t enough — beauty is only one part of the formula. 

I tried using software like Midjourney and produced some gorgeous raw outputs simply by prompting. But it didn’t satisfy me. 

‍AE: What does the human need to add to the raw output to produce something that connects with an audience?

‍G: My process isn’t very rational. I just keep experimenting until the image resonates with what I want to communicate. I’ve always had a talent for drawing but it always frustrated me because I would tend to look for perfection. Now, finally, I’ve been able to give up this idea that the end image should be perfect. That only worked for me when I was doing it for someone else, because the boundaries were really strict. I used to make physical models for museums, films, and advertising photography, building insects and dinosaurs with high quality plastics, epoxy, polyester, vinyl, and acrylic.

When digital photography came along, I had already been working with Photoshop and 3D for four years, so moving to digital photography and then AI was easy for me. But, with painting, you have an empty piece of paper and you can do anything imaginable — that always paralysed me. 

‍AE: As the notion of the artist as a hybrid becomes increasingly popular, we are starting to see new, less anthropocentric, and often nonvisual, approaches that seem to reflect a posthuman condition. Like others who work with virtual worlds, figures are often evacuated from your landscapes. However, your recent works generate an extraordinary hybrid species. Can you explain this development?

‍G: I’m fascinated by the fine line between figuration and abstraction. For me, one of the best qualities of these AI systems is that the machine has no intelligence at all — you just feed it images. It doesn’t know the difference between a photograph and a painting. For humans, these are two distinctive disciplines but the system doesn’t distinguish. Indeed, if you train it on both photographs and paintings, you can produce a photograph-painting. I once used Artbreeder to generate a Volkswagen-banana. [Laughs]

As an output, it looked like a convincing photograph, but it was also something in between. Parts of your brain register realism, but this was simultaneously completely abstract.

‍AE: One of the historical problems with algorithms is that they’re trained on images that are labelled wrongly. Has that ever invaded your workflow?‍

G: Not really, because I’m not looking to synthesize existing images, I’m looking to find everything in the middle and on the fringe. I want to explore the unknown, and anything that the algorithm does wrong is something I can use. At one point, I asked the coder of BigGAN’s visual interface if the model had been trained on the work of Beksiński, Frazetta, or Picasso, and he said: “no, that’s all in your mind. That’s what you want to see and that’s what you recognize, but it has nothing to do with these [artists].” 

This latent space is so vast and flexible that you can find almost anything that exists in your mind in this huge forest of weird shapes, including a new species.

‍AE: In works like Latent Space Folded (2021) the figurative elements seem to exist underneath a kind of epithelium as an interconnected network of beings, almost dissolving into one another. I’d love to see that on a very large scale. 

‍G: I really like that my work forces the viewer to use their fantasy. Anything that looks too finished or too obvious is not really interesting art to me. Francis Bacon’s stated aim was to deliver a “visual shock” to the viewer.¹ You need to move people with what you do because there is so much out there.

A couple of years ago, I was at a crossroads in my life, and I asked myself: “what is my true nature? What do I really want to be?” But the idea that everything had been done already took my enthusiasm away. By contrast, AI is directly connected to our present time. And, in 2022, there’s still a lot to be discovered. I need to feel like an explorer. When I was young, if I needed information on something, I had to go to the library, or to a bookshop. Now, with the Internet, I can type in any subject and receive information about it — too much information. In the past, If I wanted to see art, I needed to go to a museum or buy a book. Now I just type in an artist’s name and I can see practically everything they ever made. 

Today, the unexplored is everything in between — everything that isn’t visible. What I’m doing right now is something that isn’t found in a search engine.

‍AE: For the show, “Liebe Maschine, male mir” (“Dear Machine, paint for me”) — a collaboration between Kate Vass Galerie and Elementum — your project, SOMNIVM (2022), is shown alongside works by artists of different generations who also use autonomous systems. How does it feel to be juxtaposed with figures like Hein Gravenhorst, Espen Kluge, and Frieder Nake, who cover such a spectrum of practices?

‍G: I feel like the odd man out here, because I am probably the only one who doesn’t know a line of code. But, then again, I don’t think that’s the most important thing — it could even turn into some kind of fetishism.

I really love showing physical works. On the other hand, I just went to the Van Gogh Museum in Amsterdam and literally every painting was behind glass. You can’t even approach the paintings because there are barriers to keep you away. But everything you see can be downloaded in high resolution with a 4K screen and calibrated colors. These days, it’s almost like the digital equivalent is even more true than what you see in real life. I know that behind the glass is actual paint on canvas, but they won’t really let me look at that. 

There’s a whole generation growing up now for whom the digital world is maybe even more important than the physical world. On the other hand, most of us on this planet have been brought up without digital art. So I love it when something works as a print. For this show, we are using thermal sublimation techniques whereby the works are baked into a very thin layer of aluminum. They look gorgeous in real life.

‍AE: When people buy an NFT, they are buying a token that is often linked to the media only insecurely. In this climate, the transactability of digital art almost takes precedence over its display. But your work has a lot of detail that invites exploration. How important is display to you?

‍G: Having that feeling of owning a print that you can put on your wall is another form of fetishism. It’s not necessarily better than seeing something on a screen. We want to own things, and I think prints are a way to satisfy that need. But, if I take my nine-year-old son as an example, he uses tokens to buy funny hats in a video game — they can be much more important to him than something that he keeps in his room.

‍AE: You speak in a way that embraces your work’s commodity form, which makes sense to me because you’re also a prolific NFT collector. What’s the best way for me to experience a work by Ganbrood?

‍G: Any way you want. I’m now making postcards for every event I go to and I give them away. The works also look great on a screen. I’d love to experiment a little bit more — to do a machine-woven tapestry, for instance. That could really work for my pieces. 

The truth is, it doesn’t really matter how the work is displayed. The question is whether you have an easier connection to something on your wall or on your phone. 

For this new show, I worked really hard to make something that worked as a whole, with a common theme. All the works are loosely based on a work by Johannes Kepler called Somnium (“The Dream,”)(1634), which is often regarded as the first sci-fi novel. Kepler was both an astronomer and an astrologist, which sounds ridiculous today. But, at the time, it was completely normal to do both things. I like that dualism — for one person to have different roles. I think that my work is both magical and at the same time based on the latest technology. 

‍AE: Your latest dreamscapes are less obviously indebted to the Western canon of fine art. Some of the works recall early Japanese painting as well as the designs of Syd Mead. What are your aesthetic touch points right now?

‍G: Of course, someone like Van Gogh was heavily influenced by Japanese prints. I also see their influence on European comic art of the 1980s. For me, AI is a great way to explore these similarities, and where they overlap and touch each other. I love the aesthetics of Ukiyo-e — the simplicity of the clear lines and color palettes. 

‍Jean Giraud (better known as ​​Mœbius) also had a huge impact on me growing up. I adored his series, The Incal and The Airtight Garage, which were crazy fantasy stories. The most beautiful and influential comic that I’ve ever read was Little Nemo in Slumberland by Winsor McCay. He was clearly inspired by Asian art as well as Russian artists like Bilibin. Everything is tied together. I also use thangkas and mandalas in my work, which clearly examine geometrical forms and spaces at the same time. Fairy tales and mythologies are based on symbols that we recite in our minds. They always return to mine.