I loveThe Expanseseries, and one of the really impressive things about the books is how consistent they are. One way is in their quality; the five books I’ve read so far have all been quite good. But they’re consistent in other ways, too. All but one is in the 500-page range. They all have a blurb from George R.R. Martin somewhere on the front or back cover. And they all feature evocative art from Daniel Dociu.
That art, which adorned the covers for all nine books from Leviathan Wakes, published in 2011, to Leviathan Falls, published in 2021, is indelibly tied to James S.A. Corey’s series. But, if The Expanse was starting its run now instead of 13 years ago, it’s difficult to imagine publisher Hachette Book Group or co-authors Daniel Abraham and Ty Franck opting to go with a style that this closely resembles the look we now closely associate with AI-generated content.
I started reading the Expanse novels when AI art was still in its DALL-E Mini infancy, at the end of 2022. At the time, I thought the covers were interesting because they were so exact in some ways, and so fuzzy in others. Leviathan Wakes shows what appears to be a spaceship next to what seems to be an asteroid, but you would be hard pressed to get more specific than that. The second book, Caliban’s War, is much harder to make out. In the foreground, there’s what looks likea satellite or an escape pod. But the background is vague. It could either be a planet or a massive metal structure, a la the Death Star.
The other covers are similar. You can tell that something science fiction-y is happening but, until you read the book, you likely won’t be able to make much sense of it. The covers look great, sell the vibe of the books, and do an effective job of unifying a decade’s worth of fiction, even though you can’t quite tell what they are.
The Expanse artist Daniel Dociu may look familiar to video game fans. That’s because he served as the face model for Father Grigori inHalf-Life 2.
But this art style has been effectively marginalized by the rise of AI-generated images, at least for the time being. AI-created content is defined by a similar balance. The things that matter most tend to be hyper-detailed and presented in sharp focus. But, as your eye wanders away from that focus point, you may notice that the rest of the image is less carefully rendered. You’ll rarely see an AI picture of a beautiful woman with three eyes, because her face is the focus. But, you may notice that she has too many fingers, or an extra leg, or that the hem of her shirt ends at an impossible angle.
The Expanse book covers don’t get details wrong, but the art follows a similar pattern, with some objects in sharp, detailed focus, and others blurred or blocky. The big difference is that a human artist made the decisions about what should be in focus and what should be blurred. The result is expressionistic, rendering the feeling of space opera through a blend of sharp lines, overlapping colors, and bursts of light. AI could patch together bits of stolen work in accordance with those criteria, but it couldn’t match the fine balance that Dociu achieves. Like the Protomolecule in the series itself, it can replicate, but cannot truly create.
