Hello, Followers of Fear. There are sixty days till Halloween. What have you done to prepare?
Artwork produced by artificial intelligence, or AI art, has become kind of a thing as the technology has advanced. In fact, some months back, there was an app that became a fad for horror writers to generate images and even book covers. And earlier this week, John Oliver featured it on his TV show, Last Week Tonight with John Oliver. Here’s the video, if you’re interested. It’s hysterical.
John Oliver’s wife and children must have such a blast telling people who don’t know who he is what he does for a living. “Oh, my husband/dad? He sexually harasses Adam Driver and marries cabbages on TV once a week.”
Anyway, it got me interested enough to want to play around with AI art, so I made an account with one of the recommended companies, Midjourney, and went to work. The results were not only fascinating, but gave me some thoughts on the nascent AI art industry.
First, here are some of my successes. As it turns out, Midjourney’s AI program does very well with Lovecraftian/cosmic horror entities. For example, here’s Cthulhu destroying Las Vegas because I’m not a big Vegas fan (though under the right circumstances I’d visit again).
Then, in order, we have Yog-Sothoth, Shub-niggurath, Nyarlathotep, Azathoth, two pictures of the Deep Ones, the Color from Outer Space, and the King in Yellow.
And portraits of famous people tended to go well with the program. Here are Stephen King, Anne Rice, and HP Lovecraft, three of my biggest influences as a writer, as well as one of Lizzie Borden with an axe. Because why not?
And because I love ballet and try to put dancers in my stories when I can, I made a series of ballet posts, with their titles in the captions.
Pretty cool, right? Most of them look really good. However, those were success stories, like I said. Here’s what happened when I tried to create Jason Voorhees from the Friday the 13th films.
Yeah, I know. None of those come close to looking like Jason! He’s arguably more famous than Cthulhu, but these were the results. And then there was my attempt at Freddy Kreuger from A Nightmare on Elm Street. Oy freaking vey!
Yeah, who is that? Not Freddy Kreuger, who is definitely more famous than Cthulhu! Honestly, it looks more like my conception of Leland Gaunt, the antagonist from Needful Things by Stephen King. How could the AI get our favorite subconscious serial killer so wrong?
And those are just a few examples. I went through several failures trying to get anywhere close to a specific idea or image, only to give up when I realized the program just couldn’t create it. I couldn’t even create a decent Sailor Moon or an accurate tiger image, and both those prompts have enough references out there that they should have been easy to create something accurate.
Even some of the successes took a lot of work and experimentation, such as The King in Yellow or Little Dancer and Anubis. I won’t even go into detail about how hard it was to get a good Alice in Wonderland pic that wasn’t too trippy. I mean, I know the source material is plenty weird, but not all art based on it has to be super-surreal!
And while I call some of them successes, they aren’t perfect. You can especially see it in the ballerina series, where features like the head and limbs look odd or bent in ways you wouldn’t see on a normal dancer or a Degas painting. Only the tutus and bodices come out well. The rest of their bodies can be a mixed bag.
It kind of reminds me of that scene in 1986’s The Fly, where Jeff Goldblum puts steak through the teleporter and it comes out tasting weird. He figures out that the computer hasn’t learned how to move organic materials, and is instead creating an interpretation or bad imitation of what it thinks steak is. That’s what we have here: the AI has learned how to mimic and create, but it’s still leagues away from making certain things.
And honestly, I’m glad. Art is art because there’s someone behind it with a vision or a passion. You can program a computer to recreate famous art pieces or original pieces, including Batman comics and scripts or horror screenplays. And the computer might even do a good job at times. But there won’t be any passion or soul behind it. Art is art because we’re putting our love and soul into our creations.
And getting to work with a creator? That’s even more special. Believe me, I’ve done it before, and it’s amazing to see your vision come to life with their help. Especially when working on art pieces like book covers. That truly is something special.
That being said, I can see AI-generated art being used for cheap book covers as the technology improves, and the services of artists becoming more expensive. This could especially apply for small presses or independent publishers who need to lower costs while maximizing profits. The only time publishers may use a real artist is if the author is big enough to warrant it.
Hopefully that doesn’t happen, because it would mean we wouldn’t get to see some really amazing collaborations. So, for the meantime, I’ll stop with the AI art (I’ve scratched that itch and I don’t want to pay a subscription for Midjourney, anyway), and continue supporting the artists who create amazing art. Especially horror art.
And now, here are some more pieces I created. Except for a few I’m waiting to reveal till the time is right, here are my other successes. You can ask what was going through my mind when I created these in the comments below. Enjoy, and until next time, my Followers of Fear, pleasant nightmares!
Rami Ungar The Writer 