What Is the Bank System?
Long ago, Osamu Tezuka — creator of Astro Boy, the pioneering television anime — devised a method called the Bank System to produce 30-minute episodes every week.
“Bank” means to store. Unlike today, animation at that time was drawn on paper, traced onto cels, and painted by hand. Normally, cels were discarded after a production was completed. But Tezuka decided to keep and organize certain cels rather than throwing them away — bust shots, waist shots, close-ups, walking shots, flying shots, crying, laughing, lip-sync shots, and other footage likely to be reused. This was the Bank System.
Naturally, when I began Toei’s first television anime series Wolf Boy Ken, we adopted this same system.
Not every studio used it, but many did. Whenever a new storyboard was completed, a bank coordinator would search the bank for matching shots and find suitable backgrounds for reuse.
Today, original drawings and animation cels from Tezuka Productions are partially preserved and exhibited at “Anime Tokyo Station” (established 2023), a facility near Ikebukuro Station in Tokyo, open to the public.
This system was a major contribution in terms of both scheduling and budget. In the digital era, everything is stored on hard drives and can be called up as needed from folders labeled “laughing,” “crying,” and so on.
AI and the Bank System
The mechanism of AI is essentially the same as the Bank System.
One major difference is that Astro Boy’s movements could only be used in the Astro Boy series, whereas AI can transfer those movements to entirely different characters. This is what makes it groundbreaking and even more efficient.
However, this efficiency brings a problem: the codification of cuts. This connects to the question Junichi Ushiyama asked me, which I wrote about in my previous post — “Why are cuts in anime so short?”
On Codification
What is commonly called codification usually refers to the codification of facial expressions. But the codification I am concerned with is a problem rooted in cut length.
For example: a cut of someone crying is followed by a cut of the other person’s dialogue, then another cut returning to the crying person with a different expression — and so on. When short cuts continue in this way, each individual cut begins to function like a word.
To put it more precisely, imagine a four-second unit of dialogue and image repeating in succession.
There is a self-deprecating term: “one-cut animator.” It means that in a three-to-four-second piece of animation, only one sentence worth of movement is drawn alongside the dialogue. The movement is the bare minimum needed to illustrate the line. It does not connect naturally to the next movement. This is codification.
When short cuts are strung together, the sequence of cuts conveys its own message, almost like a sentence. What the creator wants to communicate may come through clearly. But on the other hand, it leaves no room for the audience to feel the atmosphere, to think, or to judge for themselves. To put it bluntly, it becomes the equivalent of listening to a one-sided lecture.
Words are symbols — cold signs with no emotion of their own. As a result, what occurs is the very opposite of the reality that artists originally seek.
The One-Cut Animator
Another problem is the decline of animators’ technical skills. When you keep doing one cut, one action, you lose the ability to animate continuous movement — actual acting.
The timing within continuous movement is itself a vital element for expressing reality. The self-mocking term “one-cut animator” described those who could no longer do this.
Toei Animation was formed by acquiring a company called Nihon Dōga — the true pioneer of anime in Japan. Its tradition was rooted in full animation, which is based on naturalism.
Many animators had firmly mastered that technique. But as they continued working in television anime, their advanced skills went unused. When occasionally asked to animate continuous acting, their technique could no longer deliver. That is what “one-cut animator” meant.
The term has since disappeared, but think of combination techniques in judo. When one throw fails, there is usually a pause before the next attempt — something you see often even at the Olympics. Executing techniques in rapid succession without pause is the combination — and it only comes naturally through constant practice.
Animation is the same. That is why I called the term “one-cut animator” self-deprecating. It was one of the merits and demerits of television anime.
How Do You Convey to AI What Words Cannot Express?
“ChatGPT” is a well-chosen name — AI can answer nearly anything that can be put into words. But for things that cannot be put into words, it can do nothing at present.
It is said that 50% to 90% of human communication consists of things that cannot be expressed in words — conveyed instead through body language, gestures, or directly through the face and body (the shoulders speak, as it were).
Unless we find non-verbal methods to communicate what words cannot explain, AI will never be able to understand these things.
AI itself honestly admits that it cannot yet do a director’s job. It does leave room for the possibility in the future. Body language could be captured through motion capture, but that would be a return to CG.
If all captured movement becomes AI data, it would increase AI’s value — which is not a bad thing.
However, I believe that unless AI acquires skin, it will never be able to direct. What matters for us humans, who do have skin, is to identify what AI cannot do. If AI excels at patterns, then we must think about what patterns can never achieve.
If today’s television — morning dramas, daytime dramas, casual dramas, anime dramas — centers on dialogue while the imagery approaches mere symbols, then these are essentially AI-ready dramas. Since that is AI’s strong suit, those are the jobs AI will take. Dialogue and voice are also among AI’s strengths, so caution is needed there as well.
Sadao Tsukioka