The History of Nintendo EAD. Part 2: The Transition to 3D

The New Console Required New Roles

When developing games for the Famicom, a small team of designers could devise the gameplay and prepare most of the graphics, while leaving the code to one of the outside companies Nintendo worked with regularly. That workflow no longer suited development for the Super Famicom. The new console displayed more colors and supported hardware scaling and background rotation. To take advantage of these capabilities, designers and programmers had to work much more closely together.

F-Zero became the first example of this new approach. Eight developers and producer Shigeru Miyamoto took a failed prototype for a Famicom Grand Prix sequel and combined it with experiments in rotating the playfield. The result was a futuristic racing game that demonstrated Mode 7 while also avoiding the problems of ordinary car animation. The vehicles hovered above the track, so the developers did not have to draw wheels or suspension movement.

F-Zero launched in Japan alongside the Super Famicom in November 1990. EAD did not turn into a huge studio overnight, but its own programmers were now working beside the designers.

A Japanese Super Famicom with a controller; this was the system that forced EAD to master new graphics features and reorganize its development process.
Pippi, edited by Alphathon / Super Famicom.png / commons.wikimedia.org

Production itself was changing at the same time. According to Takashi Tezuka, work on Super Mario World began in 1988 and took less than two years. He had often handled both level design and graphics in earlier projects, but this time he passed the visual work to Shigefumi Hino. At first glance, the change was minor. In practice, EAD was gradually moving away from a model in which a handful of generalists handled nearly everything: individual staff members were taking on specialized roles and gaining more autonomy.

Hino was the person who developed the idea of a rideable animal for Mario, something Miyamoto had been considering since the Famicom era. Early versions resembled a horse, a large lizard, or a crocodile. They gradually evolved into Yoshi, a character who would later become the center of a new movement system and a separate game.

One of the early Yoshi sketches created during the development of Super Mario World.
Nintendo / Nintendo Classic Mini: SNES developer interview – Volume 5 / nintendo.com

Eight People and Two Cars on One Screen

EAD did not immediately switch to large teams, however. Super Mario Kart was made in roughly a year by eight people, including Miyamoto. Even this small team already had two directors. EAD was still working in compact groups, but it was gradually preparing people who could take responsibility not for a single screen or drawing, but for an entire game.

The original task was simple: put two controllable cars on one screen. The prototype was later often described as “multiplayer F-Zero,” although the developers stressed that they were not trying to transfer the same game directly to split-screen. F-Zero’s high speed demanded too many resources. To support two windows, the action had to be slowed down and the racing machines replaced with small karts that were easier to follow.

At first, an unnamed man in overalls sat behind the wheel. Only three or four months later did the team try replacing him with Mario and notice that the familiar silhouette immediately made the action easier to understand. Other characters then joined him, while collisions between rivals changed from an inconvenience into one of the game’s main features. In this way, the split-screen limitation determined the speed, the vehicles, and ultimately the character of a new series.

Satoru Iwata and Shigeru Miyamoto during a discussion about the origins of Mario Kart.
Nintendo / 1. It Started With A Guy In Overalls / nintendo.com

This approach would become typical for EAD. The team did not begin with a list of characters and modes, but with a simple prototype. Could two players follow the track at the same time? Was it fun for them to interfere with each other? Did the vehicle still feel good at a lower speed? Often, familiar characters and visual themes were added later, after the team had determined that the core mechanic was engaging on its own.

A Portable Zelda Grew Out of an “After-School Club”

EAD did not work exclusively on home consoles. After A Link to the Past, several employees began porting the game to the Game Boy. Tezuka later compared the work to an after-school club: the developers were learning new tools without much pressure and initially had no intention of making another major Zelda game.

SRD programmers took part in the experiment, and Kensuke Tanabe and Yoshiaki Koizumi later joined them. Once it became clear that the small system could support an independent adventure, the port idea was abandoned. Unused concepts from A Link to the Past became Koholint Island, the story of the sleeping Wind Fish, and an unusual world that did not have to follow Hyrule’s rules too strictly.

That freedom can be seen even in the details. The game included characters and creatures reminiscent of those from Mario and Kirby, while conversations with the islanders became more important than in earlier entries. The Game Boy’s small screen did not prevent storytelling, but it did force the developers to make the world more compact and densely detailed, giving each location its own character. The experiment was released in Japan in June 1993 as The Legend of Zelda: Link’s Awakening.

Link’s Awakening showed that even an unofficial experiment could grow into a full-fledged game. For the developers, this kind of work was also an opportunity to take on unfamiliar tasks. Koizumi, for example, worked on the story and learned unfamiliar tools on his own during the project—experience he would soon draw on in later games.

Polygons Required a Processor Inside the Cartridge

EAD could not make its next technological leap alone. Miyamoto wanted to create a game built entirely from polygons, but the standard Super Famicom was not powerful enough. A solution was found together with the British company Argonaut Software: roughly two years were spent developing a special processor later named Super FX. The cartridge became more than a storage medium—it became a hardware expansion for the console.

This was how Star Fox came into being. Miyamoto handled the game concept and production, Takaya Imamura designed the characters and overall visual identity, and Tsuyoshi Watanabe created the polygonal models. The British specialists shared their experience with 3D programming and helped build development tools. Through this collaboration, EAD quickly learned a technology that would soon stop being exotic.

Even Super FX did not remove the strict limitations. Only a small number of vertices could be displayed on screen, so ships, buildings, and enemies had to be assembled from extremely simple shapes. Repeating patterns suggested surface detail, while the aircraft’s route was restricted to control the number of objects in view. Star Fox’s distinctive look appeared not despite the weak hardware, but because the developers consistently designed around its capabilities.

Development artwork of vehicles from Star Fox, whose shapes had to be simplified to fit the limited polygon budget.
Nintendo / Nintendo Classic Mini: SNES developer interview – Volume 1 / nintendo.com

The project gave EAD not only a new series, but also practical experience with three-dimensional graphics. Even more important was the habit of quickly building a working version, testing it on the console itself, and removing anything that hurt performance or clarity without hesitation. By the Nintendo 64 era, this approach had become essential.

Yoshi’s Island Chose Illustration Over Fashionable Graphics

After Super Mario World, Shigefumi Hino began receiving responsibility not only for graphics, but also for planning and direction. Not every prototype he worked on reached release. When the team decided that the familiar Mario platformer formula had run its course, Yoshi offered a way to build the game differently: slow the pace, add egg throwing and a flutter jump, and make Baby Mario part of an unusual damage system in which he could be lost and then recovered in time.

The new hero also required a different visual style. The developers wanted the game to feel like a child’s drawing. A young Takashi Nogami scanned an image made with markers and spent about two weeks experimenting with how to process it. At the time, the pre-rendered graphics of Donkey Kong Country were making a strong impression, but the team chose not to imitate them. The deliberately hand-drawn look made Yoshi’s Island feel not more technologically advanced, but more alive.

This was also when a weakness in EAD’s organization became visible. Tezuka and Hideki Konno were moved to Super Mario 64, and the new project later drew away other specialists as well. For a time, only three designers and the SRD programmers remained on Yoshi’s Island. Staff eventually returned and completed the game. The episode clearly shows how EAD worked in the mid-1990s: teams had not yet become independent studios, key employees were moved between projects, and the same technical partners supported several groups at once.

This system caused delays, but experience did not remain trapped within a single series. Hino moved from graphics into direction, Konno carried knowledge about racing and multiplayer modes from one project to another, and Koizumi combined story work with 3D animation. EAD grew not so much by creating new departments as through people who took on more responsibility with every new project.

Super Mario 64: Control Mattered More Than Scenery

The move to the Nintendo 64 forced the developers to rethink even the most familiar actions. In a two-dimensional game, the direction of movement and the distance to an obstacle are almost always visible. In three-dimensional space, the player must understand at the same time which way the character is facing, how fast he is running, where the goal is, and what the camera is doing.

Miyamoto returned to the director’s role for Super Mario 64. By then, Koizumi had been teaching himself 3D graphics on an Amiga computer and showed his experiments to Miyamoto. In the new project, he worked on Mario’s model and animation, but his responsibilities quickly expanded. The camera, enemies, and environment were all tied to the hero’s movement, so they could not be developed separately.

The Nintendo 64 controller’s analog 3D Stick could register not only direction but also the degree of tilt. This allowed Mario to walk, run, and turn at different speeds. Before building levels and designing objectives, the developers had to make controlling Mario enjoyable in its own right—even if the player was doing nothing more than running, jumping, sliding, climbing, or swimming. Only then did they begin building the rest of the game around his movement.

The Nintendo 64 controller with its central analog 3D Stick, whose capabilities shaped the controls of Super Mario 64.
DreCube / Nintendo 64 controller.jpg / commons.wikimedia.org

The camera became a separate part of the controls. It had to place itself in a useful position without distracting the player. EAD never found a universal solution. In Super Mario 64, the angle could be adjusted with buttons, while in later games the developers repeatedly had to decide how much freedom to give the camera and when to control it automatically.

Work on Super Mario 64 also affected neighboring projects. Staff from the Mario Kart 64 team were reassigned to Miyamoto, so the whole division helped finish the racing game. Its developers found their own compromise: the tracks were made in 3D, while the racers remained flat images that always turned toward the camera. This saved enough memory to display eight competitors at once and preserve the four-player mode.

Shigeru Miyamoto during a discussion about the development of Mario Kart and the history of the series.
Nintendo / 1. It Started With A Guy In Overalls / nintendo.com

Super Mario 64 launched in Japan alongside the Nintendo 64 in the summer of 1996. For EAD, this was more than a successful move of Mario into 3D. The division gained a set of tested solutions for animation, control, camera behavior, and the structure of a polygonal world. These solutions could now be studied, changed, and adapted to completely different genres.

Ocarina of Time Required Five Directors

Work on what would become Ocarina of Time began with a small team. Toru Osawa and Jin Ikeda handled it at first, and Koizumi joined them after Super Mario 64 was completed. The earliest experiments focused on sword fighting. The team even considered a first-person view: a detailed hero model might require as many resources as the surrounding environment, while an invisible Link would simplify the problem.

Koizumi insisted that the hero had to remain visible. A third-person view, however, created two new problems: the camera could lose sight of an enemy, and the player had difficulty aiming an attack precisely in three-dimensional space. The solution was Z-targeting, a system that locked onto a selected target and coordinated the movement of Link and the camera.

The developers found a clue during a visit to Toei’s historical film set in Kyoto. In a staged fight, several enemies surrounded the hero but attacked one at a time while the others waited. This theatrical technique showed how to make a group battle understandable. The participants later remembered some details of the visit differently, but the central principle carried over into the game: targeting identified the active enemy, while the others attacked in turn.

Link and Navi in The Legend of Zelda: Ocarina of Time; the visible hero, camera, and selected target had to work as a single system.
Nintendo / ゼルダの伝説 時のオカリナ / nintendo.co.jp

The team drew another lesson from Mario 64. Koizumi compiled a list of problems the team had encountered during Super Mario 64 and tried not to demand difficult jumps or other precise movements from the player. Link therefore jumped across small gaps automatically, allowing the player to focus on choosing items, finding routes, and solving problems.

As the game grew, almost the entire Nintendo Information Development division, which included EAD, became involved. The work was divided among five directors. Osawa coordinated the project and handled the script, while the other directors took responsibility for individual systems, events, the camera, and characters. SRD’s programmers handled the technical implementation of these systems. This was no longer a small group in which one director held the whole game in his head, but a large team with several closely connected areas of work.

Ocarina of Time was released in Japan on November 21, 1998. By then, EAD knew how to expand a team without giving up on prototypes. The developers first tested sword fighting, then the camera and targeting, and only afterward began building a large world around rules that already worked. Not every Nintendo 64 experiment, however, could be completed in its original form.

The 64DD Failure Turned an RPG into Animal Forest

The 64DD disk add-on promised what ordinary Nintendo 64 cartridges lacked: a large amount of rewritable storage and a built-in real-time clock. Producer Katsuya Eguchi suggested using these features to create a “communication field,” a world that several family members could enter at different times, change in some way, and then notice the traces left by one another.

The 64DD disk drive installed beneath a Nintendo 64; Animal Forest was originally developed for this hardware.
Evan-Amos / 64DD-Attached.jpg / commons.wikimedia.org

At first, the game looked nothing like the familiar Animal Crossing. The developers planned a large adventure with dungeons and four islands corresponding to the seasons. The player commanded animals with different abilities, while another family member could later continue the shared journey. The central idea was not a specific RPG system, but the connection between players: they did not have to sit in front of the television at the same time to feel each other’s presence.

Because of delays to the 64DD, the project had to be moved to a standard Nintendo 64 cartridge. Only one megabit remained for save data. The team abandoned the four islands, the dungeons, and almost the entire adventure. A clock had to be built directly into the cartridge, and the enormous world was reduced to a single small village.

Normally, such a list of losses would sound like a description of failure. Here, however, the restrictions helped the team understand what kind of game it was actually making. If family members were no longer going to complete an adventure together, they could instead settle in one place: buy furniture, decorate rooms, send letters, and notice changes after someone else had visited. The currency, Tom Nook’s shop, and home decoration were added around the preserved idea of communication.

The animals changed along with the concept. Originally, they walked on four legs and served as companions with special abilities. When the adventure gave way to conversation, the characters were placed on two legs. They could now look the player in the face, hold objects, and live nearby. The animals stopped being part of an RPG system and became villagers with personalities of their own.

Promotional artwork for the original Animal Forest on Nintendo 64: the player and the villagers are gathered inside a furnished house.
Nintendo / どうぶつの森 / nintendo.co.jp

Animal Forest was released in Japan on April 14, 2001, when the Nintendo 64 generation was already nearing its end. The 64DD itself failed to meet expectations, but its features led EAD to think about real time, persistent changes, and indirect communication between players. Moving the project to a cartridge meant abandoning the original adventure, but the idea of indirect communication between players survived and became the foundation of Animal Forest.

Other 64DD experiments met a similar fate. Mario Artist: Talent Studio allowed players to create a character, including one based on a photo from the Game Boy Camera, but the developers struggled to understand what purpose such a double should serve. The idea returned in the GameCube project Stage Debut and again failed to find the right use. Only much later did the simple Mii avatars find a natural role in Wii Sports. At EAD, a cancelled project often became a storehouse of ideas that had not yet found the right application.

EAD’s Most Important Legacy Was Its Development Method

In a single decade, EAD went from a team that was only beginning to program games on its own to a division capable of running several three-dimensional projects at the same time. But small groups were not simply replaced by large ones. Super Mario Kart grew from an experiment with two cars, Link’s Awakening from an unofficial port of an existing game, Star Fox from a partnership with the developers of a new processor, and Ocarina of Time from a small sword-fighting prototype.

The same principle appeared again and again. First, the developers built the simplest working prototype and tested the core mechanic. Then they chose the character, technology, and team that suited it. When conditions changed, as they did with Animal Forest, everything unnecessary could be discarded while the central idea remained.

This approach did not eliminate chaos. People were pulled from one project into another, much depended on SRD and outside partners, and some ideas waited more than ten years for a suitable system. Yet the accumulated experience did not remain within a single team. The camera work in Mario 64 helped with Zelda, the limitations of multiplayer racing shaped Mario Kart, and an experiment for the 64DD became the foundation of a series that continued on later consoles.

By the early 2000s, EAD mattered to Nintendo for more than Mario and Link. Technical limitations suggested new mechanics, while ideas from unfinished prototypes were reused in later projects. The same approach underlay even EAD’s very different games.