Cover image: a corrected key animation drawing from Neon Genesis Evangelion by Norio Matsumoto and Shunji Suzuki
This article is part of the History of the Kanada school series
Many factors contributed to the progressive decline of the Kanada style in the late 80s and onwards onto the 90s. The simple explanation is that it was probably growing stale, and spectators started getting tired of it, which in turn led producers to favor works with a different aesthetic. But the new conditions of the industry in the 90s also played a part. The storyboards and layouts of now-experienced directors with a vision started getting more detailed, closer to a movie’s production, even on TV series and OVAs. Animation directors themselves took more and more importance as they started correcting not only the key frames, but also the layouts that were made by the key animators.
This, in turn, only incited animators to search for new, less idiosyncratic, ways to express themselves. On the other hand, more and more new animators and directors started entering the field, people who had learned the craft during the Kanada school’s golden days, but had also been impressed by the high-budget, realistic movies of the late 80s, from Akira to Venus Wars. They, too, sought to change the status quo.
More than the Kanada school, whose members appeared throughout the 80s, this new group of animators deserves the word “generation”. The six major animators I will focus on were all born in the second half of the 60s, and debuted as key animators within the same five years: Atsuko Ishida (1986), Masayuki Kobayashi (1988), Atsushi Wakabayashi and Norio Matsumoto (1989), Tetsuya Nishio (1990) and Hirofumi Suzuki (1991). Most of them have since risen to prominence as some of the most important animators alive, and left a deep mark on 90’s and 2000’s animation. This was made possible by the new style they all contributed to establish, one I would term “flow animation”.
I already mentioned the concept in previous articles, notably when I talked about one of Shinji Hashimoto’s sequences on The Hakkenden, but now is the time to look at it in depth. It was, as the origins of some of its animators will show, the synthesis between the Kanada style and the realism of the late 80s, and it focused mostly on character animation. From the Kanada school, it took the playfulness and systematic use of framerate modulation, and from the realists, it borrowed a sense of weight, detail, and nuanced expression. What it added were three things: a priority put on fluidity, a disregard for rigid character designs in favor of motion, and a generalized use of the smear, which slowly overtook the impact frame as the queen of anime techniques. If I call it “flow” animation it is because, in contrast to the angularity of late Kanada-style animation and to the sculptural volume of Takashi Nakamura-inspired realism, its emphasis on motion creates an almost liquid impression, as if the characters were made not of flesh but rather of a shapeless, ceaselessly moving kind of matter.The key actor in that transition was, I believe, Mitsuo Iso. He did not immediately come into contact with the new generation I mentioned, and when its members started making themselves noticed, Iso had only started his own long and rich career. But I think it was Iso who, the first, definitely broke with the Kanada school’s approach to animation, conceived as a series of poses, to center on motion itself. More than Iso individually, it was Gosenzosama Banbanzai’s groundbreaking role in anime history which was probably the cause of it all.
From starting point to turning point
Before going into what the flow animators worked on, it’s interesting to know where they came from. The eldest member of the group, Atsuko Ishida, started as an in-betweener in Kaname Pro, the household name of the Kanada school, and was a student of Mutsumi Inomata: some of her first work was on Plawres Sanshirô, Birth, and Dream Hunter Rem. It is most probably alongside Inomata and on Birth that she first got into contact with flowing, cartoony and expressive character acting. This was the way the Kanada style was moving towards back then, spurred on by Kanada himself and by his most talented student, Shinsaku Kôzuma. Ishida then quickly got into contact with the realists, most notably with her first key animation on Gundam ZZ and Megazone 23 Part II in 1986. On the other side of the spectrum, Norio Matsumoto was a thoroughbred realist: he started at Oh Production, did his first in-betweens on Robot Carnival, Grave of the Fireflies and Akira, and started as key animator on Takashi Nakamura’s Peter Pan’s Adventure.
It was, however, on Ranma ½, and especially its second (Japanese) season, Nettô-hen, that most of them met: it’s there that Ishida and Matsumoto, as well as Hirofumi Suzuki, Tokuyuki Matsutake and Masayuki Kobayashi did some of their first notable KA work. If their own talent could bloom in such a way, it was probably thanks to the support of Atsuko Nakajima.
A regular on Rumiko Takahashi adaptations, she had started in-betweening, and then went on to key animating on Urusei Yatsura from episode 142 onwards. On the last stretch of the series, she quickly became a key member of the team and was promoted to animation director on episode 192. After that, she went on to be animation director for many episodes of Maison Ikkoku, and finally became character designer and chief animation director of Ranma, where she supervised many of the youngsters’ episodes. Her slightly angular, animation-friendly designs probably played some part, as did the more frequent use of smears pioneered in Maison Ikkoku’s comedy and Ranma’s emphasis on action and acting.
While Ranma is not comparable with Urusei Yatsura in terms of quality and influence, both shows can certainly stand an analogy, in that they both opened their decade with a staff of some young, talented animators whose work prefigured most of the aesthetic of the next ten years. The first third of Nettô-hen, until episodes 44-50 (late 1990), was dominated by Kobayashi and Matsumoto who then probably left to work on The Hakkenden and Roujin Z. Suzuki and Matsutake started working on it from episode 101 (late 1991) to the end of the show (late 1992); of the group, Ishida is the only one who animated on it from start to finish.

The early episodes, i.e. the Ishida-Kobayashi-Matsumoto ones, offer very interesting chances to see what stayed from the Kanada style, and what changed. As this sequence shows, speed lines and impact frames were still on the agenda, but they were repurposed in order to support the proliferation of smears. For example, in this frame, it’s virtually impossible to distinguish between the smears and the speed lines—that’s a use of the technique that even Shinsaku Kôzuma, the main smears user of the Kanada school, didn’t make. It seems, instead, to have been inspired by Iso’s use of the technique on Gosenzosama Banbanzai.
While this particular way of drawing them reminds me more of Masayuki Kobayashi, the grandmaster of smears was, from the start, Norio Matsumoto, who would use them in different manners at the same time. For example, in this sequence, there are two distinctly different kinds of smears. In the first shot, the red and black smears, interspersed among speed lines, seem to exist on their own, floating in the air, as pure afterimages of the punches raining down. But on the fourth shot, as Ukyiô jumps down and swings her giant spatula, the smears are attached to the object, and appear as dents on the straight lines. Another characteristic of Norio’s style was animating almost systematically on the 1s and 2s.
Ishida appears to be an intermediary between the two, but it is clear from her work here that she was a born character animator and designer. Her acting is rich and detailed, but where she really shone was in the character expressions, always very cartoony and expressive. These have an illustrative quality that would perfectly find its place in the more angular character designs of the late 90’s—a trend to which she would contribute significantly with her work on the Brave series, Magic Knight Rayearth and Shamanic Princess, but which would also develop in her illustrations and manga. On the second part of the show, Hirofumi Suzuki and Tokuyuki Matsutake formed a true duo, for multiple reasons: for both of them, Netto-hen was their first work as key animators, and they always worked on the same episodes. Moreover, they were clearly taught by Nakajima, and influenced by what Matsumoto had already done. For example, while the space between frames is wider than Matsumoto’s (there was less mastery and ambition), this sequence is a pure copy of his style in its intricate and fluid movement and in the shape and use of smears. Generally, their work was less impressive, but it represents the first examples of the newfound success of flow animation. Moreover, it was the start of a long collaboration between the duo and their inspiration: they would keep working together on Gunsmith Cats, You’re Under Arrest, Medarots, and of course Naruto, where the “flow animation” style reached its peak and could rightfully deserve to be called the “Matsumoto style”.
Old meets new
Ranma was indeed a turning point, but it was still too small to really make waves in the industry as a whole. However, things were clearly changing, even beyond Ranma: OVAs like The Hakkenden and Sukeban Deka were symptoms of a wider turn. Around the years 1993–1994, a double shift occurred, definitely confirming flow animation’s new dominance in anime aesthetics. On the one hand, the style found wider and more critical success, while on the other, in an apparently more unexpected change, experienced Kanada-school animators seemed to join the new style. This happened on two series: Irresponsible Captain Tylor, in 1993, and Yû Yû Hakusho, between 1992 and 1995.
Tylor is perhaps the most representative work of the transition from the 80s to the 90s. In its writing, it initiated a new trend that would express itself with even more success and talent in 1995 with the darker Neon Genesis Evangelion and in 1996 with the more light-hearted Martian Successor Nadesico: taking a look back at the history of mecha and science-fiction, and at once both turning it on its head and paying homage to it. In terms of staff, it brought in some of the great names of the 80s, who would go on to leave their mark on anime in the years to come.
The director, Kôichi Mashimo, is mostly remembered for his “girls with guns” trilogy in the 2000s, but he had already directed emblematic 80s movies such as Ai City and Dirty Pair: Project Eden. From Project Eden, he brought with him Atsuko Nakajima’s husband, Tomohiro Hirata. While Tylor was Hirata’s first work as character designer (and his female characters were probably a big influence on Nakajima’s and Ishida’s 90s designs), he was already a veteran animator: his career had started on Galaxy Express 999, and he had gone on to Urusei Yatsura, for which he was a regular key animator from episode 44 onwards. However, he would not be chief animation director on the series: there was none, as expected from Mashimo who always sought to give freedom and independence to animators.
In terms of animators, the show brought back together two ex-Graviton members: Kôji Itô, who was also joint mecha designer, and Shôichi Masuo. Their style doesn’t always come out obviously, except maybe in the near-perfect episodes 22 and 23, which feature some exemplary spaceship and effects animation. However, their collaboration was made under the patronage of perhaps the Kanada school’s most talented member: Shinsaku Kôzuma. Maybe as a throwback to their previous successful works as a group, they animated the cel-animation parts of Tylor’s opening together. They truly shared the work as people who knew each other would have: Itô and Masuo worked on all the mechanical and effects parts, while Kôzuma did all the character animation by himself—and ended up, it must be said, under the spotlight.
His style had always been unique, even among Kanada-school animators, but his work on Tylor is the confirmation of an evolution that would bring him even further from the Kanada style and closer to flow animation: his shapes became more round than they had ever been, and he started putting much more attention to the shading in an ever-increasing search for fluidity. But the real highlight in his Tylor OP is probably the timing, which is more irregular than ever, constantly oscillating between 1s, 2s and 3s. Kôzuma did little work on the series, and it is seldom very flashy (except the wonderfully surrealistic dream scene he did on episode 19), but it is always full of care and details: his unique motion plays a large role in Tylor’s unique personality, expressed in his wide walk cycles and irregular, bursting movements.
Considering this, it’s hard not to think of Matsumoto’s spectacular contributions to the series as a sort of answer to Kôzuma. This is especially the case in his solo second ED which features yet another iconic walk cycle that would inspire a later great character animator, Ryôma Ebata. Besides that, he animated on nine episodes, and while you can find moments relatively consistent with his work on Ranma, it’s also notable that he started playing with slower timings (like this sequence, mostly on 3s and with wide spacings) and went for complete realism in some other scenes that exhibit a perfect mastery of volume, depth, and nuanced expressions. With Tylor, Matsumoto confirmed his ascension to the status of master animator.
At the same time, just as Tylor opened a new trend of parodic space operas, action shônen was going through one of its masterpieces: Yû Yû Hakusho, which started airing in November of 1992 and ran until January 1995. It was produced by Studio Pierrot, and its staff mostly featured one household name, whom I’ve already considered at length for his work on Urusei Yatsura: Motosuke Takahashi, the studio’s “trump card”. After Urusei Yatsura, he most notably directed (and was animation director for) three other Rumiko Takahashi adaptations between 1985 and 1987—Fire Tripper, The Supergal and Laughing Target—as well as some forgotten, if not forgettable, movies and OVAs: Aitsu to Lullaby (1987) and Harbour Light Story (1988). Considering that, with Yû Yû Hakusho, Ninkû (1995–1996) and later Naruto (2002–2007), Pierrot played a central part in the shift of action animation from the Kanada to the Matsumoto style, Takahashi’s role in that transition is probably instrumental. His influence, as well as remains of the 80s aesthetics, are everywhere in the early part of the show.
The most obvious example of this lies in the show’s character designs: initially in the hands of Minoru Yamasawa, they ended up being given to Mari Kitayama and Masaya Onishi from episode 20 onwards. If Yamasawa’s designs still look very 80s, it is even more the case when they were under Takahashi’s supervision as animation director: see, for example, the difference in Keiko’s face between episode 4, with Takahashi as animation director, and episode 6, with Atsushi Wakabayashi at the same position. Wakabayashi’s drawing is much more angular and feels more adult than the childish and round expressions apparently favored by Takahashi.

The latter’s influence is not only visible in the drawings themselves, but also in the key role he played in training and supporting a young episode director who made his debut on Yû Yû Hakusho and would go on to leave a deep mark in the following decade: Akiyuki Shinbô. Until then, Shinbô had only done some unremarkable key animation work. He had started as an in-betweener on Urusei Yatsura, and it was probably thanks to his senpai, Takahashi, that he became episode director on this new series for no fewer than nineteen episodes, including some of the most important ones.
On those nineteen, Takahashi was storyboarder or animation director for seven. And it does seem that, initially, Takahashi’s style and personality were dominant: the most striking example is episode 19, where Takahashi the storyboarder was surrounded by a couple of youngsters, Shinbô on direction and Wakabayashi on animation direction. However, apart from some smears here and there that testify to the slow shift at hand, both the direction and animation are very conservative, as they would be during the largest part of the show: fairly static, if not awkward, animation, only supported by a frequent use of speed lines and impact frames. It would not be until the show’s “Dark Tournament Arc” that the animation would really take off.
However, even before that, the show started fostering the next generation. Beyond Shinbô, the other person who benefited from an unexpected and sudden promotion was Atsushi Wakabayashi, who had only been key animating for three years when he became animation director on the series for twelve episodes, most of them with Shinbô and/or Takahashi. Wakabayashi had studied animation with Norio Matsumoto, but they briefly went their separate ways when Wakabayashi joined Pierrot and Matsumoto Oh Production. In an interview, Wakabayashi confessed that, back in animation school, he had never understood his classmates’ obsession with Yoshinori Kanada; his first real discovery was Takashi Nakamura’s episodes of Gold Lightan, a taste that clearly links him to the realist lineage.
As for Wakabayashi’s actual animation, he quickly followed in Matsumoto’s footsteps: the influence is obvious, especially in the way he used smears and walked the path of even more flowing animation. The other standout of the early part of the show, also clearly moving in the same direction, was Tetsuya Nishio, whose work here was just a prelude of what was to come.
After these humble beginnings, the Dark Tournament Arc was a turning point, mostly because it was full of fighting and special abilities, and because the animators seem to have been granted a lot of freedom. Covering it all in sufficient detail would warrant an entire separate article, so I will just mention a few standout scenes. On episode 47, Wakabayashi delivered one of the show’s most iconic sequences, a pure moment of flow animation which seems to owe more to Shinji Hashimoto and Shinsaku Kôzuma than to Matsumoto. The shape of Yusuke’s body is simplified to the extreme and reduced to just lines and circles. Although the modulation is very stark (sometimes jumping directly from 4s to 1s), the timing is often very slow (mostly on 3s and 4s) and the space between poses is as extreme as can be. In a way, you could almost say that Yusuke’s movements in the first shot are barely animated because there are so few in-betweens—and yet it feels so fluid because the shapes and poses are so dynamic by themselves.

Nishio, on the other hand, made some of his best work on episode 49. The use of smears and detail in the cloth and character animation here still betray Iso and Matsumoto’s influence, but by then he had already grown out of simple copying. What’s notable here is that, whereas Matsumoto and company were still very much only doing character animation, and would keep doing so for a while, you see Nishio trying out effects, not only in the explosion and smoke effects, which are not exceptional, but also in the animation of the flying ghost figures. They are drawn in such a way that you can’t really determine what they’re made of: is it some sort of smoke, or liquid-like substance? The creatures sometimes look like large bands of cloth that are waved around and move irregularly and unpredictably. A lot of work is put in the texture and makes this cut feel at the same time unique and representative of how effects animation would evolve once the dominance of Kanada-style effects was over.
But the real achievement of the series was, of course, episode 58, one of the visually most impressive episodes of the decade. The staff was probably a match made in heaven: the Shinbô-Wakabayashi duo was at its height, and among the animators you could find more youngsters who would become key staff of Pierrot’s and flow animation’s future great shônen action shows. Besides Wakabayashi himself, you had Atsuko Inoue (one of the only persons Wakabayashi has said equaled Matsumoto’s skill) and Masayuki Yoshihara (future director of Uchôten Kazoku). But, most importantly, Shinbô had managed to bring in a legend: his old friend and teacher Shinsaku Kôzuma. It was not his first involvement with the show, since he had already storyboarded episode 54, a role he rarely took on. Maybe I’m overcrediting Kôzuma here, but I’d like to believe that he, Shinbô and Wakabayashi reached an agreement over what the episode should look like: a meeting and synthesis of the young and the old, where the Kanada style would bow out in a last great explosion to leave the room for the younger, more dynamic flow animation.All the conservative and boring elements of the animation were taken out, most notably the tired speed lines and repetitive impact frames, in favor of a strong colour direction and stark lighting contrasts. The shading was very heavy, a call back to the late 80s aesthetic reminiscent, for example, of Shin’ya Ohira’s early work and so-called “wakame shading”. On the other hand, all the animators, especially Yoshihara, favored very round and soft designs, which were nevertheless very different in their simplicity from the lolicon and bishôjo sensibilities of the 80s.
More generally, the division would be as follows: all the effects would be in a bold and renewed Kanada style, while the character animation would completely follow the search for flow and fluidity. The use of Kanada dragons, colour and flashes makes the reference obvious: this was an attempt to replicate and overcome Kanada’s most iconic work, the finale to Genma Taisen. And it worked perfectly. Matsumoto’s followers didn’t have an effects animation of their own, and because of that, they were as yet unable wholly to rewrite anime’s visual vocabulary as Kanada had. But this episode showed that, with competent (not to say visionary) direction, they could very well rely on Kanada-school effects. This was not without consequences: at the same time, it kept the Kanada style from becoming irrelevant, but also relegated it to effects, thus helping to obscure the central contributions it had made to character animation.
The year anime changed
But all this didn’t happen instantaneously, just because of a single episode, however revolutionary. The traditional Kanada style had yet to be completely buried—and this was very much the case after one of anime history’s most important years: 1995. This was, of course, the year Neon Genesis Evangelion aired, starting in October. I won’t go over how much the series turned the entire world of anime upside down: both a very classical and experimental mecha, its stellar writing and direction set a standard throughout the decade’s following half. But one thing people tend to forget, because of the show’s complex production history, is that it also featured some exceptional animation and an awe-inspiring animators lineup. In this respect Evangelion very much confirmed anime’s visual trends since the decade’s start.
Just like romantic comedies and action shônen, mecha was also starting to break away from the Kanada style in the early 90s: Gundam entries such as Stardust Memory and Victory are two very different attempts to find a new aesthetic distinct from that of the 80s. Evangelion was an interesting case, because Gainax, and especially Hideaki Anno, had never truly adopted the Kanada style. However, the show featured some of Kanada’s old collaborators, who worked closer to his style, in its staff: Shôichi Masuo (who animated on seven episodes) and Kôji Ito (who worked only on episode 10). The veteran Kôzuma was also brought in on episode 16, but his work clearly had nothing Kanadaesque to it anymore.
Besides that, Evangelion was yet another step in the wider recognition of the younger generation of animators, inside and outside of Gainax. The outsiders didn’t go as far in the “flow” direction as in Yû Yû Hakusho, but their work still left a strong impression. The most important figure in that regard was of course Norio Matsumoto, who did some key animation on episodes 5 and 8. But there was also another: a rising star of mecha animation from Sunrise who made himself even more well-known by working on this popular series, named Yutaka Nakamura. While he only worked on 2 episodes, and alongside such heavyweights as Masuo or Masayuki, his work immediately stood out thanks to its fluidity and attention to detail. In this sequence, even though Nakamura had just a short part (roughly from 0:40 to 1:00) and many stills, everything is very powerful and manages to convey the impact in a much more sober way than Keisuke Watanabe and his flashier Kanada-style effects.
Finally, even before End of Evangelion, the TV series offered a playground to one of the greatest animators of the time: Mitsuo Iso, who animated on two key episodes, 1 and 19, and delivered some outstanding work in both effects and mecha (character?) animation. But this was just one among many displays of talent for Iso: more than the moment Evangelion came out, 1995 was the year the realist school confirmed its place at the forefront of animation by creating two of its masterpieces, Junkers Come Here and Ghost in the Shell. The two movies are very different, and, Iso apart, they barely had any staff in common. But, considering they came out the same year, it’s hard not to take them together when looking at the realists’ development.
Junkers was the third in-house production of the small outsourcing studio Triangle House, which had been created in 1987 and would close down in 2000. The project was initially an ambitious one, and was given to Shin’ya Ohira, who invited many great names both from and outside the realist school: Iso, but also Osamu Tanabe, and legendary Madhouse animator Manabu Ohashi. But Ohira’s pilot was a bit too experimental—and, most notably, it took much more time and money to make than expected. The producers were not pleased and decided to put the movie into other hands. The director, Jun’ichi Satô, had already made a name for himself with the first season of Sailor Moon, but the real change was in the new animation director, Kazuo Komatsubara.
Komatsubara had been a character designer and animation director since the early 70s, and was the co-founder of one of the most important outsourcing studios in anime history, Oh Production. Oh Pro worked with all the most important anime studios: TMS, Tôei, Madhouse and Ghibli. Thanks to this, Komatsubara had done character designs and/or animation direction for Kanada many, many times—he was even brought on Martian Successor Nadesico to animate, alongside Kanada and Kazuhiro Ochi, some of the scenes from the fictional super robot show Gekiganger 3 in their 70s style. Although Komatsubara was by all means a living legend, bringing him on the movie can only be read as a conservative move: putting an established and experienced figure on the movie would have been a guarantee that it would not try to be as challenging as Ohira’s pilot had been.
The movie did, however, feature excellent animation work—Tanabe and Iso, especially, were the stars of the show, while Ohira had more or less left the project. It must also be noted how much Satô’s direction was instrumental in highlighting the animation. The simple storyboards were always made in such a way to have multiple actions on screen, allowing the different characters to interact in the frame and the animators to give the full measure of their talent.
Just like for Triangle Staff, Ghost in the Shell was the first major work of Production I.G., which had also been created in 1987—but the scale was completely different. The film adapted an already-popular manga; its direction was given to the visionary Mamoru Oshii, who completely made the story his own, while the legend Shôji Kawamori handled the mechanical designs. If Junkers’ animators represented the free-spirited and polyvalent side of the realist school, with Nakamura, Ohira, Hashimoto and Iso, Ghost in the Shell was in the hands of two very different personalities: Hiroyuki Okiura and Toshiyuki Inoue. In stark contrast with the simplicity of Komatsubara’s designs, which allowed the animators to create a rich and nuanced range of expressions, Okiura’s designs were photorealistic, big, heavy and blessed with a strong sense of volume. The realism of Junkers gained from a sense of abstraction, whereas that in Ghost in the Shell was radically embodied, in a way that perfectly fit the movie’s themes.
With those two movies, realism had found for itself a comfortable place: feature films, often ambitious like Ghost in the Shell, or at least comfortable enough environments to foster the demanding nature of those animators. They would mostly leave TV animation to their students, who, like Matsumoto, had started going in a different, but complementary direction: from acting to action, from nuance and weight to fluidity and speed. In such a context, Kanada-style animation was unwelcome outside of niche OVA productions, and it risked remaining what it was in Yû Yû Hakusho: a few techniques such as speed lines and impact frames. Simple as these were, they fitted the complex and angular character designs of the 90s—but they were starting to recur without originality. If Gainax and Evangelion had merely been confirming the dominance of the new flow animators, the Kanada style would have remained a memory of the 80s, a peripheral kind of ornamentation. But it was also fostering a new generation who would rekindle the Kanada style’s initial energy: the one who would be known as the “neo-Kanada school”.