Style, Surface, Substance, and Slop

Recently, OpenAI released a new image generator that touts the ability to render images in the style of Studio Ghibli’s films. I’m not going to link to it, nor am I going to embed any images of its product. It’s not something worth engaging with for a multitude of reasons. It would be best ignored if there weren’t valuable lessons to be learned from its folly. One can easily identify the problems this generator has. For one, it conflates the oeuvre of Hayao Miyazaki with the entirety of Studio Ghibli. For another, and this is the more prudent issue, it removes all context and intentionality from Miyazaki’s art. The content and substance of his work is erased in favour of pure surface-level imagery that you can apply to literally anything. A picture of yourself? Of course, easy. Popular internet memes? You got it. A white supremacist rally? Nothing is off-limits here, no matter how disrespectful it is to Miyazaki’s legacy, artistic integrity more broadly, or basic human decency.

I consider a few reasons why someone might find this generative AI (note: the term “AI” is used very loosely here as the services offered by OpenAI do not resemble intelligence, human or otherwise, in any meaningful way) image filter to be interesting. One possibility is that it’s something akin to a fidget spinner, a hollow diversion that can catch your eye and occupy your thoughts for a few minutes. While this sounds patronizing, it’s the perspective I’m most sympathetic towards. “Press the button for an almost interesting image” is relatively innocuous in the abstract, and it’s hard to explain to someone why it’s aesthetically or morally wrong. The more sinister explanation is that this is a Revenge of the Nerds scenario in which a bunch of Silicon Valley tech bros are wielding this technology against artists out of spite. They know artists hate this stuff, Hayao Miyazaki in particular, so why not do something with the express purpose of pissing them off? Generative AI evangelists are sharing these images not despite their offensive nature, but for that very reason. Of course, there is a third possibility, one that is simple without being generous: AI tech bros actually think this is what art is. For someone who is at best ignorant of the arts and at worst resentful of them, it’s easy to imagine that art is simply about developing a style and then applying said style to anything or nothing in particular. Not only do they disregard the actual skills needed to find one’s style, they altogether throw out the idea of any sort of conceptualization of art. Art is just images you look at. What the image depicts and how it is depicted is utterly irrelevant to them.

The relevance of this to video games might not seem apparent until you remember that the video game industry and video game culture alike are dominated by tech bros and businessmen. That’s not to say there aren’t meaningful artists making games even in the mainstream, nor that video game critics have nothing worthwhile to say. But the dominant discourse, the zeitgeist of video game aesthetics, is that a video game’s art style is just a coat of candy-coloured paint layered on top of the game, and that this surface level aesthetic is wholly unrelated to the rest of the game. Worse still, “art style” is treated as a binary value. It’s either there or it isn’t. Generally, the absence of art style is referred to as the dreaded realism. In an ironic twist, realism is considered to be both prohibitively expensive and the natural state of things before an art style is applied.

All of this is to say that art is not only irrelevant to the rest of the game, but irrelevant to itself. One looks at any game that isn’t made to resemble a cartoon and declares “there’s no art style, it’s just realistic,” to which I ask, realistic what? The objects, textures, lights, and effects that make up a game do not just exist from creatio ex nihilo, handed down as the natural order of things. Artists are responsible for creating and placing these assets just as much as they would be in an unrealistic game. One may point to the use of photogrammetry, motion capture, etc. as evidence to the contrary, but these are merely tools for the creation of the assets. Artists are still required to decide the what of the matter, what objects are being scanned into the game. Yet in the name of art, the artists responsible are so often left out of the conversation. By the same token, that being that conceptualization has complete supremacy over execution, the artists are excluded for unrealistic art styles as well. Just as in the Studio Ghibli generator, the act of creation is entirely divorced from the act of imagination.

AI evangelists claim that their technology democratizes art. One no longer needs to study, practice, or even think. All of that is done for you, making the average philistine just as capable of being an artist as someone who has spent years honing their craft (bizarrely at odds with Silicon Valley’s usual meritocratic libertarian leanings). A similar argument appears often in video games: unrealistic, stylized art is superior because it’s easier to make. This is, of course, a slap in the face to the artists who actually make those games. Claiming that theirs is easier work erases the technique and craft required to achieve what some perceive as a simpler style, and it’s a peculiar concession to the opposing crowd that thinks anything less than photorealism is worthless. Indeed, one can see this in practice. A good comparison is between Dragon Ball FighterZ by Arc System Works and the upcoming Hunter X Hunter: Nen X Impact from Eighting. Both of these are adapted from manga published in Weekly Shonen Jump, one by Akira Toriyama and the other by Yoshihiro Togashi, and while their styles aren’t identical, one can identify clear similarities. Yet Dragon Ball FighterZ is still a sight to behold seven years after release, and Hunter X Hunter: Nen X Impact appears woefully undercooked. “Art style” can’t explain this disparity. There are considerable differences in budget between these two games, but that just serves to prove that unrealistic graphics are not automatically easy to create.

Another way of looking at the matter is that OpenAI is associating their technology with the famously affable and agreeable films of Studio Ghibli as a means of whitewashing their business. Please do not look at the environmental degradation, copyright infringement, or dubious finances behind this industry. Instead, look at the cute anime characters. Unsurprisingly, one can find a similar argument in video games: realistic graphics are prohibitively expensive and require large amounts of human labour, thereby perpetuating the industry’s crunch and layoffs. Bright and unrealistic graphics are simple and easy, thereby immune to these problems. This argument falls apart under a moment of scrutiny. Fortnite, the cultural juggernaut created by Epic Games, is a heavily stylized, cartoony-looking game, yet it has been reported that its creation required months of crunch time.

OpenAI’s appropriation of Studio Ghibli serves to illustrate how our culture has a distorted perception of art, and this perception is strongest in video games. We cannot let differing forms of artistic expression be simplified into a wallpaper that can be slapped anywhere without purpose, and this requires consideration and appreciation of both the conceptualization and technique behind these styles. These cannot exist in a vacuum, however, and viewing one form as more pure or direct only serves to delegitimize art more broadly.