The good news is, I’m a journalist, so I went ahead and asked several artists, researchers, and art critics about the aesthetics of AI art. First, I called Amelia Winger-Bearskin, an artist and professor at the University of Florida. Winger-Bearskin has been Cataloging different visual trends she’s noticed in recent AI art. She calls a trend Nightmare Corporation—usually illustrated with images created by Google’s Deep Dream, an older generator released in 2015. It specializes in swirling, psychedelic, memory-like images from a particular acid trip. pain. “Definitely a Prog-rock influence,” she said. Another category Winger-Bearskin explores, which she calls Dada 3D, sounds a lot like the goofy scenes I refer to playing around with these generators. She describes it as “something like a surreal living room game.”
Aside from categorical trends, Winger-Bearskin has noticed broader styles in these generators. She considers Disney-style Western animation and anime to be obvious influences, as does the tendency to view whites as the default race—she suspects, as a result, training these creators on animation-heavy dataset, Disney-style western anime. , and pictures of white people.
Lev Manovich is also very interested. The cultural theorist and professor at the City University of New York has been lurking on Midjourney’s Discord server since last year, analyzing how people use generators. After Midjourney released an update last fall, he saw some changes in what people were pushing the generator to do. For example, after it became better at representing people realistically, the requirements for portraits of both men and women increased.
Digital artist Sam King begins to follow the AI art scene closely for the first time in 2021. Excited by what they see, they begin To share their favorite images on social media, building a curator following as the technology evolves. They describe the earlier wave of generators as favoring “the abstract, nimble things”. (These generators are known as general adversarial networks, or GANs. I have seen one few Everyone call this look, rather uncreative, GANism.)
King considers the latest wave of generators, known as the diffuser model, to be stylistically distinct. Just as oil paintings and watercolors produce distinctly different effects, GAN generators and diffusers create distinctly different images. If you want a more realistic rendering, such as Tony Soprano drinking a cappuccino with Shrek, then diffuse models are more likely to produce convincing results. “In theory, you could create all sorts of different aesthetics with these machines,” they said. However, more practicality does not necessarily mean more stylistic diversity. Like Winger-Bearskin, King sees Disney influences and frequently cropped cartoons, as well as comic book art.
“The rhetoric from these companies is that you can do anything you can imagine. It’s about this wide open frontier. But, of course, popular culture follows specific stereotypes and jokes,” Manovich said. He sees variations on several themes over and over: “Fantasy, fairy tales, comic books, video games.”
