How The Legend of Zelda: Ocarina of Time Was Made

It Began Not with Hyrule, but with a Sword

The first experiments with a 3D Zelda began even before the Nintendo 64.

Shigeru Miyamoto and Yoshiaki Koizumi experimented on the Super Famicom with a polygonal version of Zelda II: The Adventure of Link. The game remained two-dimensional and showed the hero from the side, but they wanted to build Link himself out of polygons.

The idea never became a full project. It did, however, leave the developers convinced that sword fighting in the next Zelda should be much more expressive.

In 1995, Takao Shimizu prepared a short demonstration for the future Nintendo 64. It showed a polygonal Link fighting an armored opponent.

The clip lasted only eleven seconds and was not yet part of an actual game. It simply showed what Zelda might look like on the new console.

Shimizu soon moved on to direct Star Fox 64, while Toru Osawa took over the Zelda project. Together with Jin Ikeda, who had recently joined Nintendo, he began putting together the future Zelda 64.

At first, only two people were effectively working on the game.

There was no story, no map of Hyrule, and none of the familiar dungeons. There was only a direction: make a Zelda game in which sword fighting was fun.

When development of Super Mario 64 neared completion, Koizumi joined the team. He worked on Link's model and movement, the camera, 3D environments, items, and in-game events.

That small experiment gradually began turning into Nintendo's largest project of the time.

What Zelda Learned from Super Mario 64

Ocarina of Time is often described as a game built on the Super Mario 64 engine.

That is not entirely wrong, but it can make it sound as though Nintendo took a finished technology, replaced Mario with Link, and started building Hyrule.

The reality was more complicated.

What connected the two games most of all were people, tools, and accumulated experience. Koizumi moved over from the Mario 64 team, Takumi Kawagoe had worked on its camera, and the programmers at SRD already knew how to make the new console render a 3D world.

Zelda's earliest work did rely on that technical foundation. Over time, however, its code and overall structure changed so much that the developers regarded it as a separate system.

One of the early Zelda 64 prototypes, with enemies and interface elements unlike those in the final game.
Nintendo Power Source / Zelda 64, April 21, 1997 / glitterberri.com

Mario 64 answered the question of how to run and jump freely through three-dimensional space.

Zelda faced different tasks. Link had to approach characters, read signs, push objects, shoot, use a sword, and fight several opponents at once.

The camera had to show not only the path ahead, but also the target with which the player was interacting.

Even reading an ordinary sign became a problem. Link could be standing next to it while facing slightly away, leaving the player circling the right spot and trying to find the correct position.

In Mario 64, a poor camera angle might be merely annoying. In a game built around combat and interaction with the world, the same problem made the controls unreliable.

The team also lacked ready-made tools. The programmers wanted designers to be able to place enemies and assemble dungeons themselves instead of asking a coder to handle every small change.

The team therefore had to create some of Zelda's editors from scratch—the programs in which designers assembled game spaces. For the dungeons, it prepared around fifty reusable objects that could be selected, configured, and placed in different rooms without help from a programmer.

The same pattern continued throughout development: Mario 64 gave the team a technical foundation and experience, but no ready-made answers. The camera, controls, world interactions, and development tools still required solutions of their own.

Zelda Could Have Been a First-Person Game

While the developers still did not know how large a world the Nintendo 64 could handle, Miyamoto considered a fairly radical option: limiting the game to Ganondorf's castle.

Different rooms in the castle would have led to separate worlds, much like the paintings in Princess Peach's castle in Super Mario 64. One door might open onto a gloomy meadow, another onto the sea, and a third onto another self-contained adventure.

In the worst case, Link would never have gone outside at all.

This was not the final concept, but a fallback in case a complete Hyrule proved too large for the console.

Another early plan called for a first-person view. Players would travel through Link's eyes, while the hero himself would appear on screen only during combat.

This approach would have simplified the camera and allowed more resources to be spent on environments and enemies.

Koizumi did not like the idea. He had already made Link's model and did not want the character he had worked so hard on to remain invisible for most of the game.

Once Link's appearance and movement became expressive enough, the first-person idea was abandoned.

Link stayed on screen, and the team still had to solve the camera and performance problems.

Early versions also had a dedicated jump button. It was later replaced by automatic jumps from ledges.

Miyamoto worried that manual jumping would make players measure every step and turn Zelda into another platform game. The designers therefore set the jump points and directions while building each level: when Link reached the right edge at the correct angle, the game determined where and how he should jump.

What Really Happened with the 64DD

Zelda 64 was originally developed for the 64DD, a disk-drive expansion installed beneath the Nintendo 64.

The 64DD, the disk-drive expansion for which Zelda 64 was originally being developed.
The Jedi, edited by Evan-Amos / Nintendo-64DD-expansion / commons.wikimedia.org

Its magnetic disks could not only store data but also record changes. Miyamoto imagined a world that remembered the player's actions. Link's footprints, for example, might remain on the ground after he walked across it.

The disks had one serious disadvantage: data was read from them more slowly.

The necessary data could be retrieved from a cartridge almost immediately, while a disk drive had to seek it on magnetic media. For a game with an enormous number of movements, that became critical.

By the end of development, Link had around five hundred basic animations. The game combined upper- and lower-body movements to produce roughly a thousand variations: the hero could run while drawing a weapon, turning, or raising his shield.

Loading everything into RAM in advance was impossible. Constantly waiting for the disk to find the next movement was not an option either.

According to the team's recollections, Koizumi said outright that he could not make “his Link” move properly on the 64DD. Nintendo ultimately chose a standard cartridge.

Another version of the story sometimes appears: that Ocarina of Time was moved to cartridge because the 64DD failed commercially.

The timeline does not support that explanation.

By late 1997, Miyamoto was already saying that the main game would appear on cartridge and that a separate 64DD expansion would follow later. The device itself did not go on sale in Japan until December 1999, more than a year after Ocarina of Time.

The 64DD's commercial failure did kill many of Nintendo's later plans. It could not, however, have caused the earlier decision to move the main game.

After release, the team continued work on Ura Zelda, a version with altered dungeons intended for the 64DD. It eventually reached players on the GameCube under the title Master Quest.

Ura Zelda should therefore not be confused with the early disk-based version of Ocarina of Time itself. They belonged to different stages of development.

The Ninja Show That Changed 3D Games

Abandoning the first-person view left the team with its main problem: how could it show Link, his opponent, and the space around them at the same time?

Osawa, Koizumi, and Ikeda went looking for ideas at Toei Kyoto Studio Park, a film studio and theme park in Kyoto where historical movies were shot and staged sword fights were performed.

It was a hot day, so the developers stepped into a cool hall where a ninja show was taking place.

On stage, several opponents surrounded a samurai. One attacked with a chain-and-sickle weapon known as a kusarigama. The samurai caught the chain, it pulled taut, and the attacker began moving around him in a circle.

Toei Kyoto Studio Park, where the developers saw the staged fight that inspired the structure of Z-targeting.
663highland / 130706 Toei Kyoto Studio Park Kyoto Japan04s3 / commons.wikimedia.org

Osawa saw the chain as an invisible connection between Link and an enemy.

With such a connection in place, moving the stick sideways could mean not an ordinary step, but circling the selected opponent. Link would always know where to look, while the camera kept both characters in view.

Koizumi noticed something else.

A whole crowd was attacking the hero, but the enemies did not rush him at the same time. It was a choreographed performance: one attacked while the others waited their turn.

The same principle was transferred to the game. When the player selects a target, the other enemies hold back and wait their turn. They can move nearby and create a sense of danger without turning the fight into a chaotic pileup.

Ocarina of Time could therefore create the feeling of “one against many” combat without taking control of the situation away from the player.

Z-targeting did more than help players strike an enemy accurately. It brought the controls, camera, and enemy behavior together into a single system.

And it all began because three Nintendo employees decided to escape the heat by watching a ninja show.

How a Triangular Marker Became Navi

All that remained was to show which enemy the player had selected.

At first, a simple inverted triangle appeared above the target's head. Koizumi was not satisfied with such a technical symbol. Since the team was making Zelda, the marker could become part of its world.

The triangle was replaced by a fairy.

Koizumi imagined the fairy as a tiny girl, but the Nintendo 64 could not display such a small character with enough detail. She was therefore simplified into a glowing orb with wings, and the system was named the Fairy Navigation System.

Toru Osawa shortened the word navigation into the name Navi.

Once the marker became Navi, it gained new functions. The color of its glow indicated whether the figure before Link was a friend or an enemy. The fairy could offer advice, take part in dialogue, and guide the hero through the story.

Navi also gained a role in the story: Link meets his fairy, travels with her, and eventually says goodbye.

Navi even helped solve a performance problem in Kokiri Forest. The console could not display every villager among all the trees at the same time.

The developers first showed the villagers' small glowing fairies, then loaded their owners as Link came closer. Players saw a light and understood in advance that someone was there.

The hint system was not perfect, however. After release, Miyamoto called Navi's advice one of the game's weakest elements: the system often repeated the same tips and could not determine what was actually troubling the player.

Inventing a glowing fairy was easier than teaching her to give good advice.

Navi could not simply be removed. Without her, the controls, part of the story, and several technical tricks would all have fallen apart.

Navi was not the only familiar part of Ocarina of Time that emerged during development. The same was true of young Link.

At First, the Game Had Only Adult Link

Today, childhood and adulthood seem fundamental to Ocarina of Time.

For a long time, however, there was no journey seven years into the future. The game had only adult Link.

For a project that began with the idea of dramatic sword fights, this seemed logical: an adult hero could wield a large sword and reach bigger enemies.

About a year and a half before release, Miyamoto and other staff wanted to bring back young Link. The team liked the more familiar and endearing version of the hero from the older games.

Simply shrinking the character was not enough. The game had to explain why two versions of Link existed in the same story and how Hyrule changed between their eras.

Time travel became the answer.

Link draws the Master Sword, travels seven years into the future, and can then return to childhood. A decision made late in development—to bring young Link back—thus grew into the central theme of the entire game.

For the writer, this meant rewriting the story; for the designers, checking items and environments in two eras; and for Koizumi, almost doubling the work on the model and its movements.

Working diagrams of Hyrule for child and adult Link. Aonuma suggested looking for differences between the two versions of the world.
Denfaminicogamer / まず2Dゲームで開発、社員300人で1週間遊ぶ!? 新作ゼルダ、任天堂の驚愕の開発手法に迫る。「時オカ」企画書も公開！ 〖ゲームの企画書：任天堂・青沼英二×スクエニ・藤澤仁〗 / news.denfaminicogamer.jp

The team found a clever shortcut. They reduced the adult model according to specific proportions and reused some completed animations.

Scaling alone was not enough: the walk, other movements, and interactions with objects still had to be redone. The workload therefore increased not twofold, but by roughly fifty percent.

At first, the developers planned to prevent players from returning to childhood immediately. That would have left missed events inaccessible for several hours.

The story was rewritten again, and Link was allowed to move freely between the two eras.

In Ocarina of Time, story, technology, and animation constantly affected one another. It was almost impossible to change one element without disturbing everything else.

Five Directors Instead of One

As the project grew, one director could no longer be responsible for everything. Different parts of the game developed in parallel, so Ocarina of Time was led by several directors at once.

Toru Osawa was responsible for the story and coordinated the overall direction. Yoichi Yamada handled game systems. Yoshiaki Koizumi led the 3D systems, camera, and creation of Link. Toshio Iwawaki of SRD headed programming.

The fifth director, Eiji Aonuma, was the last to join the project.

Until then, he had worked mainly with outside studios and had repeatedly asked Miyamoto to transfer him to an internal Nintendo team. When the Zelda developers needed more people, Aonuma was invited to help.

By that point, most of the story already existed, and the team was actively creating dungeons, enemies, and other parts of the game world.

Aonuma designed six dungeons from the early and middle parts of the game, created most of its enemies, and worked on boss battles.

From left to right: Toshio Iwawaki, Eiji Aonuma, Takumi Kawagoe, Yoshiaki Koizumi, and Toru Osawa, key members of the Ocarina of Time development team during a 2011 interview.
Nintendo / 1. 「運命を変えたソフト」 / nintendo.co.jp

Some of his working documents have survived. In a 2017 interview, Aonuma brought hand-drawn dungeon layouts and a map showing the connections between areas of Hyrule.

Among them was a draft of the dungeon inside Jabu-Jabu. Notes on the page included phrases such as “tip of the tentacle” and “the entire room is soft and pliable.”

Aonuma himself laughed at how casually everything had been written down. Yet these sketches and brief notes were the foundations from which complex 3D dungeons were later assembled.

A working diagram of the dungeon inside Jabu-Jabu, showing the arrangement of rooms, platforms, and tentacles.
Denfaminicogamer / まず2Dゲームで開発、社員300人で1週間遊ぶ!? 新作ゼルダ、任天堂の驚愕の開発手法に迫る。「時オカ」企画書も公開！ 〖ゲームの企画書：任天堂・青沼英二×スクエニ・藤澤仁〗 / news.denfaminicogamer.jp

Aonuma would later lead development of the next Zelda games. His role on Ocarina of Time was much narrower: he was responsible for one part of a large game alongside the other directors.

The boundaries between jobs remained loose. One employee might advise on the camera, draw storyboards, and invent gameplay mechanics at the same time.

Takumi Kawagoe, for example, joined as a camera specialist after Mario 64. He then helped create tools for cinematic scenes, began drawing storyboards, and ultimately took on much of their direction.

Miyamoto was credited as producer, but he did not limit himself to general supervision. He did not write the story alone or design every level. At first, Miyamoto allowed separate groups to experiment freely. As completion approached, he intervened more often, selected ideas, and brought the separate pieces together into a single game.

He later recalled that he had intended only to produce Zelda. Miyamoto gradually became more deeply involved in development and ultimately wrote roughly a quarter of the design documentation.

According to Miyamoto, more than fifty people worked directly on the game. Separate from them was the huge Super Mario Club testing group, with more than one hundred members.

This probably explains the different figures given for the team's total size: some count only the developers, while others include everyone who contributed to the project.

How the Ocarina Took the Place of Magic

Even the item that gave the game its title did not immediately receive its final role.

An ocarina is a small wind instrument, similar to a whistle with finger holes. Link uses one to perform short melodies, while the game's title alludes to the special Ocarina of Time.

The developers first worked out how the player would play it: controller buttons produced individual notes, while the analog stick changed their pitch and made the sound waver slightly—that is, it created vibrato.

They then had to decide why the player would perform melodies at all.

An early version had only six songs. At the same time, the team experimented with five or six magic spells, but Miyamoto found them unconvincing.

Some effects became ordinary items, while some functions were transferred to music.

The finished game contained twelve main melodies, or thirteen if the Scarecrow's Song composed by the player is included.

The ocarina changed the time of day, summoned rain, opened passages, transported Link across Hyrule, and connected him with other characters.

What might have remained an ordinary spell menu became an action performed by the player's own fingers.

Even Link's horse, Epona, was originally going to be summoned by whistling with a blade of grass. Once the ocarina became the game's central instrument, calling the horse was assigned to it as well.

All of this gave Koji Kondo an unusual limitation. The main in-game melodies had to be recognizable while using only five available notes.

He built the melodies around short motifs and tried to make each one distinct from the others.

Koji Kondo built the game's music around short melodies that could be performed with five notes.
Vincent Diamante / Koji Kondo - GDC 2007 / commons.wikimedia.org

The music of Hyrule Field was not a single long recording either. The game selected twelve different eight-bar sections and joined them in a random order.

When Link stopped, the music grew calmer. When an enemy appeared nearby, the arrangement changed and the percussion became stronger.

Players could spend hours running across the same field without really noticing that the music was adapting to their actions the entire time.

Three Years of Constant Reworking

Nintendo originally planned to release Ocarina of Time in 1997. The larger the game became, however, the further its release slipped.

Ocarina of Time went without a final shape for a long time. As development continued, the team changed not just individual details but the structure of the entire game.

Over the course of development, the team abandoned the first-person view and dedicated jump button, then added young Link, time travel, and Navi. The game moved from 64DD to a cartridge, while the adventure expanded beyond the castle to encompass all of Hyrule.

Major revisions continued even after many parts of the game were already complete.

In February 1998, the programmers changed the way the console rendered 3D graphics. Some completed models had to be discarded and rebuilt, but the game became noticeably better-looking.

When the developers changed how long Link could remain underwater, the Water Temple completely fell apart. The placement of items and the design of its rooms no longer worked, so the entire dungeon had to be rebuilt.

A working diagram of the Water Temple drawn by Eiji Aonuma in 1997.
Denfaminicogamer / まず2Dゲームで開発、社員300人で1週間遊ぶ!? 新作ゼルダ、任天堂の驚愕の開発手法に迫る。「時オカ」企画書も公開！ 〖ゲームの企画書：任天堂・青沼英二×スクエニ・藤澤仁〗 / news.denfaminicogamer.jp

Even shortly before completion, changing the Hookshot's range could suddenly send Link flying through walls.

Feedback from the Super Mario Club led the team to revise roughly thirty percent of the game.

The testers did more than hunt for bugs and check the controls. They also judged the impression made by the characters.

In the first version of the meeting with Princess Zelda, the camera lingered on a nearly motionless, expressionless heroine. One tester wrote that this princess did not make him want to set out on a journey at all.

The scene had to be redone, with new camera angles, movements, and facial expressions for Zelda.

The finished game contained more than an hour and a half of cinematic scenes. That was already the length of a short movie, even though Nintendo constantly emphasized that it was making a game.

Individual places in Hyrule were also created without a detailed plan. The basic structure of a village came first, after which artists and writers decided who lived there and what kind of house each person needed.

Hyrule Field became so large that crossing it began to feel tiring. The developers then walked through its empty areas and added enemies, hidden holes, and small events.

Hyrule was not copied from any single real place. The artists studied reference material from around the world, including books of German castles, and assembled their own world from those sources.

Miyamoto compared this process not to building a set, but to growing a small garden.

Development took more than three years. Ocarina of Time was released in Japan and North America in November 1998, and in Europe in December.

By then, players expected a great deal from the game. The Nintendo 64 badly needed a major hit, and Zelda's repeated delays only raised the bar.

Remarkably, Ocarina of Time lived up to those expectations.

A Game Built from Solutions

Ocarina of Time is often called the game that established the rules of 3D adventures.

But its creators began in a world where those rules did not yet exist, and had to invent them as they went.

Z-targeting was not created simply as an impressive new mechanic, but because Link and his opponent became lost in three-dimensional space.

Navi grew out of a target marker. Time travel was needed to unite adult and young Link in the same story. A cartridge was chosen so that hundreds of movements could load without delay. The ocarina inherited the functions of unsuccessful magic spells.

Even the vast Hyrule was not an essential part of the original concept. At one point, Nintendo seriously considered having Link spend the entire adventure inside Ganondorf's castle.

The result was a game in which almost every famous design choice grew out of a specific problem.

That may be why, so many years later, the game's design still feels so natural.