AI Art Anime Backgrounds: Creativity And The Animation Shortage
AI art will define this decade forever, and defines the way WIT, a studio that pioneers anime technologies, explored the possibilities of algorithm-driven backgrounds in a short film for a new project involving Netflix itself. The music video itself is a blend of various modern advances in the otaku cultural landscape: the singing involves an artificial intelligence (as the song is credited as (feat. AI-Code “M”), credited to the producer and musician sasakure.UK) clearly reminiscent of vocaloids; the backgrounds are firstly done by an artificial intelligence and then cleaned-up by artists behind; and at last 3D animation done by the French artist Mathias Demongeot, a concept artist for Netflix.
Though the project may hold intrinsic value as an artistical experiment, notably because of the involvement of various stylistic types of artificial intelligence applications, this has resulted in backlash coming in the way of Netflix. Globally, the advent of AI-driven art can be considered as a form of automation, and directly competes with artists on markets, ending up having sociopolitical implications. Because Netflix is the giant it is, the streaming company needs to behave according to its scale and impact on our world. Indeed, one of the consequences of the application of AI art in anime may be to intensify the current labor shortage, that already puts the artform as a whole at risk.
This is especially more true since Netflix has already claimed to support the industry in fighting against the labor shortage in prior collaboration with WIT Studio. Back then, it was done through the financing of the tuitions and living expenses of animators going through a training program done by WIT itself. Such are the words Taiki Sakurai uses to mention the labor shortage: “It is artists with extraordinary craftsmanship behind the stories that breathe life into your favorite character, cut by cut. However in Japan, animators have become a scarce profession compared to its rapidly growing demand. We want to change that.”.
May the venture into AI-driven backgrounds at ends be thought of as a new necessity, in front of a labor shortage that will not solve itself without outsourcing to other countries? The delays of airing productions like Ayakashi Triangle, Maou Gakuin and Isekai Ojisan may be associated with outsourcing to other countries, experiencing local issues like lockdown in China.
But anime itself is the result of technological progress, and the artform has never ceased to evolve and incorporate innovations upon innovations. If we try to see beyond the manichaeism of greedy corporates trying to squeeze more and cheeper productivity from creative workers, there may be wealth to be found in the use of AI art generation for anime backgrounds. Not only to cut the production costs, but perhaps for such things as streamlining or easing the ideation process. If we admit that AI art generation ends up being noisy, perhaps the noise itself can be a stylistic choice.
I once developed the idea of anime backgrounds being packed into banks and used for “homebrewed” anime productions, maybe even released into consumer markets, perhaps in a similar vein as the amateurish cultures built around Koikatsu Party or even Miku Miku Dance. And I even went as far as to predict that the repeated uses of similar stages (to use a term common to MMD fans), or backgrounds in our case, may end up feeling generic. This is one of the risks for the AI generation of backgrounds down the line, beyond the technophilic craze.
Another effect that such a background production may lead to, other than a generic feeling, that is kind of felt in some anime shows already — this is a personal feeling, but very few productions are very innovative with their backgrounds, and this is why I’m willing to commend interesting uses of background arts now — , is an uncanny valley effect. We already know that the products of AI art can be near photorealistic and still incorporate fatal flaws, for example the teeth of a jaw… And so, perhaps that people that are very sensitive to details will find themselves subconsciously cringing at incorrectness that failed to reach the consciousness of the clean-up artist.
Another thing needs to be mentioned: other productions have already made use of softwares to transform backgrounds that were photographs to begin with. And, though this creative process avoids much more the risk for copyright infringement, it is still an alternative to hand-drawn backgrounds that makes use of artificial intelligence to ease the production. Loosely related, but the styles of movies directed by Hayao Miyazaki, Satoshi Kon and Makoto Shinkai were thus imitated through the framework of AnimeGAN (Generative Adversarial Network, repository available on GitHub) back in 2020, according to an AnimeNewsNetwork article.
Through this article, we have discussed the cooperation between WIT Studio and Netflix in addressing the anime labor shortage, which also testifies to outline Netflix’s involvement with the anime industry in the early 2020s. The idea to apply technological innovations to anime is reminiscent of Yoru no Pierrot 360° in the way music videos are short, self-contained works where the poetry left to visual imagery may be proportionally greater. Here, we were dealing with AI-driven backgrounds, something which spans at least a few years back with “[a] Chinese research team from Wuhan University and Hubei University of Technology [that] developed a machine learning framework [able to] turn photographs into high-quality anime-style background art”. One of the silver linings of this adventure in the world of anime may be that more light has been shone to the craft of backgrounds, something that seems cruelly underrated to the profit of more flashiness, like with the sakuga paradigm. Subtle can be the integrations of technological progress in animation, but the future will tell us how things unfold.
