Over the past few weeks, most social media users are likely to have come across at least one post on their feeds where an AI filter is applied to an existing image to make it look “Ghibli-style,” emulating the unique aesthetic of films produced by the beloved Japanese animation company Studio Ghibli. The studio is well-known for maintaining consistent stylistic elements across almost all of its films, be it in terms of the soft coloring, the delicate character design or the richly detailed settings, all of which come together to create a warm and soothing visual palette.
However, something that may be less known about the studio is that its films are drawn almost entirely by hand, frame by frame. Behind each and every one of these frames is a series of painstakingly deliberate choices made by the artist to communicate something about the characters or setting, guided by raw, deeply felt emotions that the artist hopes to share with their audience. ChatGPT’s new AI filter attempts to reproduce all of this hard work and feeling with one click, which raises larger concerns about the implications of using generative AI for animation.
Animation is perhaps one of the greatest mediums through which a story can be told because it can bring any creative vision, no matter how unprecedented or far from reality, to astonishing life; there are hardly any limits placed on the creator’s imagination. Animators pour their heart and soul into every subtle change in the characters’ expressions and every little touch of the meticulous background detailing. Every single thing that appears in the frame comes from a unique worldview and results from hours of hard work, so it most certainly has a strong purpose to be there. The notion that a machine can instantaneously produce something of such vision and nuance is an utter failure to understand the value of the artist.
AI cannot have the intent to communicate a vision of its own; it merely reproduces what it sees and what it is fed. It takes somewhat of a cookie-cutter approach to animation, and that defeats the point of storytelling. Art should not be designed around superficial assumptions regarding what people want to see, it should be about what the artist wants the people to see. Simply sticking to existing templates only ends up losing the audience’s interest in the long term, and AI “art” is just another example of this. It may seem like a cool attraction, but the novelty effect will wear off eventually and it will perhaps become little more than a fad, much like the “old age” filter by FaceApp that was trending not too long ago.
On the other hand, Studio Ghibli films will never stop appealing to audiences. Since 2017, United States distributor GKIDS has been successfully holding its annual Ghibli Fest in theaters, where audiences turn up for screenings of both beloved classics and lesser-known gems by the studio every year. Furthermore, there is no better proof of Ghibli’s wide staying power than the box office performance of the recent 4K restoration of “Princess Mononoke” — one of the studio’s most acclaimed films — for IMAX; it earned over $4 million across North America within less than a week of its debut in late March.
The film is a stunningly potent epic that delves into ever-relevant themes of human coexistence with nature and the clash between environmentalism and capitalism. All of this is wrapped within a sweeping hero’s journey tale, yet it shuns the simplistic morals one might typically expect of such stories to delve into its layered, mature and sometimes unsettling thematic material and characters. The majestic, intricately detailed visuals paint a grand canvas that is a beautiful marriage between spectacle and realism. It is a breathtaking experience even on the small screen, but on an IMAX screen, the viewer is brought up close to every little detail serving director Hayao Miyazaki’s vision.
No amount of prompts could enable ChatGPT to ever produce something as original and poignant as the heart-stopping climax of “Princess Mononoke.” Moreover, it could never represent something as culturally specific as kodama, tree spirits that are said to reside over the forest in Japanese folklore. Miyazaki depicts each one of these dozens of spirits with such authentic and delicate attention to detail that could only come from someone for whom these spirits hold true cultural significance. Such details are inextricably tied to the powerful emotional graph of the narrative, in a way that can only be created by those who can really feel those emotions.
After all, Ghibli was never just about what the characters look like; it has always been about who these characters are and what they stand for. AI can only be a witness to emotions at best; it can never experience them, so any output it produces is merely an estimation of what emotions look like, not what they feel like. It can never allow one to tap into unique ways of feeling; it only makes one numb to the familiar ones. Thus, simply entering one’s ideas as prompts and expecting AI to fill in the rest is a fundamental failure to understand the power of animation, and of storytelling at large.
In essence, AI only knows how to produce hollow imitations of what exists, not to push boundaries. Ultimately, the Studio Ghibli style itself would not have come to exist without the unique vision and hard work of the people who created it decades ago. It has been carefully crafted in a way that could only come out of subjective, fully lived experiences and that subjectivity is the very purpose of art in the first place. Art is storytelling, and storytelling is the most innately human way that people can express themselves. If people leave it to machines to tell their stories, then what will remain of humanity?