Dan “NovySan” Novy, an assistant professor of emerging media arts in the Johnny Carson School of Theatre and Film, has created an exhibition “Infinite Hopper: An Algorithmic Journey Through Light and Space” that is on display at Sheldon Museum of Art from Jan. 21-July 13.
The exhibition uses cutting-edge generative technology to produce an endless flow of Edward Hopper-inspired imagery that never repeats or loops. Integrating works by the artist from Sheldon’s collection, “Infinite Hopper” creates a dialogue between the traditional and the contemporary, exploring how modern technology can reinterpret and honor fine art.
The exhibition came about when several members of Sheldon’s staff, including Director Susan Longhenry, saw Novy’s work at a Johnny Carson Center for Emerging Media Arts Demo Day.
“We invite everyone we possibly can so that they can see what we’re working on,” he said. “And the entire crew from Sheldon, and specifically Susan, saw some explorations I was doing with [René] Magritte because I love Magritte and surrealist artwork. She turned to me and said, ‘Can you do this with Hopper?’ And I said, sure, and started just listing off the things that might be fun.”
Sheldon owns five of Hopper’s works, including “Room in New York,” and was planning to showcase them this spring.
“We just started talking back and forth, and it evolved into a little bigger than I had sort of originally intended, which is my fault because I’m an overachiever,” he said. “So, it’s now three projections instead of just one interactive projection, mainly so I could show the different kind of methodologies that could be employed with generative machine learning and art and really any particular artists.”
On the north wall of the exhibit, viewers will see more than 10 hours of pre-rendered video using a particular animation methodology.
“I felt it was important that it wasn’t just these little things that you’d be like, I saw that one already,” he said. “You might see one repeat if you come back the next day, but maybe not. Random chance is just part of my jam. Even though I might use the same sort of description of a fictitious work by Hopper using his oeuvre and his theme because there are these random seed values that get injected into the noise field, you’ll never get the exact same image twice, which leads us to the interactive part.”
On the east wall, there is a large computer system that is running real-time stream diffusion.
“It’s running a real-time application that is both attempting to illustrate descriptions of works by Hopper, but also taking input from a ceiling-mounted camera,” he said. “That camera is looking down onto a 1930s era desktop, and there are objects that are thematically sort of similar to the works of Hopper. There are tiny train cars. There’s a bright orange book by Hemingway because Hopper was a huge Hemingway fan. There’s a green lamp. There’s a red dress. These things that you would see in a Hopper painting. And as you manipulate and move them, the application will use the color and shape of the objects themselves as additional input into the computation.”
Viewers will decide how much to interact with it.
“I’ve seen people just stay there playing with it for a good, long while, and some people really just like to come and stare and see what’s going on,” he said. “There’s definitely a playful component to it, which I love because game design is also part of my work.”
Finally, on the south wall, Novy will display more than a thousand pre-rendered and high-resolution stills of fictitious Harper paintings.
“You can look at them and go, ‘I can’t believe he didn’t actually paint this,’” Novy said. “But it’s got his style. The model has been trained on specifically every painting he’s ever done. They’re higher resolution because we have more time for them to sort of bake, so there’s more detail in them. They’re shown for one minute, and then the next one comes on. There are over a thousand of these still images, so again, the likelihood that you’re going to see a repeat is very low.”
Sheldon will also display the Hopper works that it owns.
“You can go in, and you’ll encounter this machine-learning algorithmic journey, and then you can go in and look at the real ones and come back and sort of compare and contrast,” Novy said. “It really is meant to be an exploration of contemporary and traditional together.”
Novy is a fan of Hopper’s work.
“He is one of my favorite painters for the use of light and stillness and silence,” he said. “In my way, it’s an homage to him to work within those kinds of themes and look development. For me, it’s almost a complete education in his art history. The more I work, the more I look at his work because I’m trying to figure out how to generate something similar. The more I worked, the more I noticed specific details about his work and my knowledge of him became much more deeper and much more appreciative. I’m hoping that’s what happens to the audience for this—that they can see when the animation or the interactive really nails it and gets it or when it gets it wrong. It really is educational and exploratory at the same time. And it’s meant to honor him.”
The exhibit has something for everyone, especially for other fans of Hopper’s work.
“It’s colorful. It’s delightful. If you’re a Hopper fan, you’ll definitely enjoy some of the ways we’ve been able to interpret his work,” Novy said. “There are certain short animations where we go past the diner at Nighthawks, and we go around the corner and see the cityscape around the corner. When we pull out from the pharmacy, you can see more of the city. We walk down the street that’s in Early Sunday Morning, and we see the rest of the buildings that Hopper didn’t paint because his canvas was too small. You get to see the couple in Room in New York, but you can see her playing the piano and him rifling his paper as he plays, and maybe it looks like he’s conducting along with her as she’s playing. There are all these sorts of almost theatrical-level scenes that are created. I’ve always been a fan of when animators take a work and hand-animate something.”
Novy’s expertise as an emerging media artist is critical to this work.
“This isn’t just the computer making images,” Novy said. “My background as an animator and as a visual effects artist played deeply into how these animations are created. This is my skill and experience as an animator figuring out a completely new set of tools and methodologies for doing animation. And the fact that I was able to create 10 hours of animation in only a few weeks’ time is mind-boggling.”
The generative technology is just another tool in the tool kit for artists like Novy to create work that otherwise couldn’t be done.
“I went through this exact same paradigm shift when we went from traditional animation and tradition optical chemical visual effects to what we call CGI,” he said. “When we started using computers for visual effects, there were a lot of people that bemoaned it or forecast the death of the film industry or of animation. It was like, ‘The computer is just going to do all the work now, and you’re just going to hit a button.’ But no, I still have to know how to animate. I still have to know the principles of animation and how to create the illusion of life. All these skills still come into play. I just have a tool that lets me go home at 5:00 now instead of having to be there all night. I’m not interested in creating animations that replace a human being. I’m interested in creating animations that a team of 1,000 animators could never do. That’s where I find the work really interesting.”
Novy said he’d love to do something like this again for Magritte.
“Magritte and Dali some of the other dadaists and surrealists lend themselves very naturally to this sort of machine-learning kind of work,” he said. “Their work you already sort of look at it a little sideways, and when the machine-learning algorithm gets something totally wrong, it’s often described as wow, that’s surreal. So being able to really drill down and nail what makes a Magritte a Magritte, not just visually, not just brushstrokes and his color palette and themes, but what were his ideas that were different than Dali. Being able to continually refine the prompting process so that it always gives me a great Magritte and doesn’t veer off into Dali, I would find super interesting.”