Super Mario 64 – 1996 Developer Interviews

Shigeru Miyamoto – Producer/Director

—When someone is playing Mario 64, everyone sitting around the TV gets really excited. It feels different from previous Nintendo games in that regard.

Miyamoto: Ever since Donkey Kong, it’s been our thinking that for a game to sell, it has to excite the people who are watching the player—it has to make you want to say, “hey, gimme the controller next!” We tried to do the same thing with Mario 64.

—But this time, the people who are watching seem to have a lot of fun getting involved, pointing and saying things like “move the camera over there!”

Miyamoto: An eternal theme for me with game design has been to let the players create their own vision. I don’t want to just hand players ready-made experiences—here you go, play this stage we made, solve this puzzle; rather, I want a game that allows players to try come up with their own solutions and playstyles and test them out there on the spot. I think that’s the best thing about interactivity. In that sense, I’m very happy that onlookers, too, are getting in on the creativity.

Shigeru Miyamoto in 1996.

—I don’t remember where I read it, but I believe you said once that the excitement one feels for a game doesn’t start when you pick up the controller, but earlier, when you’re walking home from school and thinking about how to get through the next part etc.

Miyamoto: Well, even if I didn’t say that, I imagine it’s true for everyone, right? Only in Mario 64, in terms of gameplay, we’ve intentionally returned to a much older feeling and style.

In the Mario games up to now, we’ve carefully crafted every stage and level down to the individual pixel. Take jumping, for example. Implementing jumping in 3D is really difficult.

In earlier Mario games, we were able to measure the number of pixels Mario could jump and know exactly what was possible. But this time, we had to design the levels so that as long as your jump was “close enough”, you’d make it; it was too hard for the player to judge. This was a design change we made in the middle of the development, when the game was far already very complete. There was a lot of booing from the staff.

—It sounds like jumping became more intuitive, and less quantitative.

Miyamoto: Exactly. But that’s the decisive difference between 2D and 3D. At the same time, it’s what accounts for the dynamism players enjoy in a 3D game. The essence of what makes a 2D game “fun” is entirely different.

—By the way, did you have any references, or anything else you relied on when creating all of Mario’s different movements?

Miyamoto: We tried out a lot of different things using motion capture, but ultimately we ended up doing it all by hand. We created a “skeleton” for Mario that was the basis of his movement.

—What is his centre of gravity? I feel like it’s in the hips…

Miyamoto: You’ve got a good eye. (laughs) The area around his hips is a big “joint” that controls which way his body moves. We created all his movements from that point of origin: when he accelerates and inclines forward, when he turns and leans left or right, etc. So Mario sort of runs like Arale-chan, with the correct sense of weight in the body.

Arale-chan, a character from Dr. Slump, appears to have inspired Mario’s movement and animation.

—But there’s a lot of unrealistic movements in Mario’s repertoire, too. Like when he does his long jump, it’s faster than if he was running!

Miyamoto: With 3D, little “lies” like that can go unnoticed. So we lied a lot! I mean, Mario is this weird old dude who can jump 3 times his height… so who’s counting? (laughs)

—The way Mario’s face moves is really great too. Like in the opening scene.

Miyamoto: That actually came from a prototype for Mario Paint 3D (that we’re still going to release). Skin animation, as it’s called, is a fairly standard thing in the world of animation, but I think this is the first time it’s actually been included in a game.

—It’s simple, but really fun.

Miyamoto: Game developers are starting to have a lot of pretensions with 3D. It used to be, in the past, people would ask me “Making games is a lot like making movies, isn’t it?” And I’d always respond, “Yeah, it kind of is. It’s a similar process.” Then they’d ask me, “So, do you want to make movies then?” And I’d always reply in the negative, that no, we don’t want to make movies—because we’re making games.

But lately, everyone making games seems to aspire to be some kind of movie director! Part of me is like, what the hell, do you all have some inferiority complex about movies?!

At the same time, when we were making the opening titles for Mario 64, we realized this was starting to feel a lot like making a movie! We decided we to try and make it interactive, something you could play with as in the rest of the game. As a title screen, if people feel it’s a little empty or lacking something, all I can say is “sorry!”

—I noticed that Mario speaks with an Italian accent in this game.

Miyamoto: There’s no particularly detailed background or anything, but yeah, it’s a given that Mario is an Italian-American from Brooklyn, New York. That voice was actually done by a professional voice actor. He did Mario’s voice five or six years ago, at a video game event. By the way, Mario talks a lot more in the American version of Mario 64. He says “Okie Dokie!” and more. Peach also speaks. We had more time before the American release, so we improved the game.

—I also noticed that there seems to be a lot fewer enemy types than previous Mario games…?

Miyamoto: There’s no special reason for that. (laughs) Normally we’d include about 80 different enemies, but this time we had just under 40. And many of them are neither friend nor foe, but something in-between. That was one of our themes for this development, actually.

It’s kind of like, we didn’t want to just throw the player into this scary world and have them go on an adventure; rather, we wanted to make a game where the player would feel, “wow, what is this mysterious place I’ve come to…” So many of the characters might look like enemies at first-glance, but they actually aren’t hostile to you. The bunnies, the penguins, the snowmen…

—I noticed Yoshi appears at the end of the game, but did you ever have plans to include Yoshi in the actual stages?

Miyamoto: There was originally an event with Yoshi. We weren’t satisfied with how it came out, though, so we removed it. But since it would be a waste not to use the model we had made, we included him there at the end.

—By the way, what happened to Luigi?

Miyamoto: Well… until February, he was in the game. (laughs) Ultimately, due to memory issues, we had to take him out. Then we were going to include him in a Mario Bros. style minigame, but because most users probably only have that one controller when they first buy their N64, for that reason (and others) we decided not to.

This image appeared in a different interview from Dengeki Super Famicom about a new improved Super FX chip in development. It appears to show a motion capture device being used to model Mario’s face (possibly made with the same prototyping computers that are referred to below) Miyamoto says they are making a “Mario Paint” style software tool as one of three titles currently in development with the new chip.

originally featured in the Super Mario 64 strategy guide

Shigeru Miyamoto – Director/Producer
Takashi Tezuka – Assistant Director
Yoshiaki Koizumi – Assistant Director
Hajime Yajima – Programmer
Yasunari Nishida – System Programmer
Yoshinori Tanimoto – System Programmer

—Mario 64 is the lead-off title for the new Nintendo 64… but how did the project first get started?

Miyamoto: Well, in the beginning… we were working on something really simple—deceptively simple, even, from the perspective of the team that would go on to finish the huge, final game. (laughs) There was a room made of simple lego-like blocks, and Mario and Luigi could run around in there, climb slopes, jump around, etc. We were trying to get the controls right with an analogue 3D stick, and once that felt smooth, we knew we were halfway there. And so, along the way, we realized wanted to create a slightly larger area for them to move around in…

—I don’t know if “slightly” is the right word there…

Miyamoto: Well, that’s how it is with all our developments. (laughs) For this game things were especially vague in the beginning, because we developed it in tandem with the N64 hardware, and we didn’t know exactly how powerful the hardware would turn out to be. That’s why, at the very beginning, we actually did our work on a large, powerful computer that simulated what we guessed the N64 hardware specs would be… then, well, we got things to a point where the controls were nice and responsive, and we thought this could be the foundation for a game. But the problem was, it had all been made on this massive computer that cost tens of thousands of dollars. No one yet believed that we’d be able to make something like this on a little 250$ machine like the N64. (laughs)

However, once the N64 prototype was finished and delivered to us, we saw that it handled the movement and controls almost perfectly. That was the moment we first realized this was going to work, so we quickly dashed off a planning spec sheet for the game. When the staff saw how long it was, they said they’d been lied to—no one told them they would have to make this massive game! (laughs) That’s how we make games at Nintendo, though: we get the fundamentals solid first, then do as much with that core concept as our time and ambition will allow.

—And in this case, that fundamental basis was the model you made with Mario and Luigi running around that room.

Miyamoto: Yes, it was being able to move Mario and Luigi around with the 3D control stick, and being able to change the camera view with the press of a button. One of our big development themes was letting the players move Mario around any way they wanted. We wanted to make a game where just moving Mario around was fun.

L-R (top): Shigeru Miyamoto (producer/director), Takashi Tezuka (assistant director), Hajime Yajima (programmer); (bottom) Yoshiaki Koizumi (assistant director), Yasunari Nishida (programmer), Yoshinori Tanimoto (programmer).

—People have described Mario 64 as “interactive animation”, and I think that term fits perfectly. Mario is truly a joy to control.

Miyamoto: That’s why I think it would have been great if we’d been able to make it two-player, with Mario and Luigi. But if we had done it wrong, it would have turned into a fighting game or something (laughs), so we’re leaving that challenge for next time.

—I really like all the nuance to Mario’s movement. It looks very natural.

Miyamoto: That was all done by Nishida and Koizumi. I should have told them to make the jump look cooler though. (laughs) About all we told them in terms of guidance was to create as many different movements as they could.

Koizumi: I didn’t think it would end up being THAT many. (laughs)

Nishida: Koizumi created the animation data, and I did the programming. I counted them all up, and there were 193 different animation patterns! And if you include the 50 or so animations that we created but ultimately rejected, it comes to nearly 250.

Koizumi: Of course, we brought a lot of that on ourselves by adding animations on our own. For instance, we’d see that Mario had no animation for pressing jump from a crouching position, so we went ahead and added it ourselves. We just kept adding more and more.

Miyamoto: Our thinking was that as players got good at the controls, they’d want to try out more and more button combinations, and if there was nothing past the basics it would be disappointing for them. So we created movements for all button combinations—of course, that means a few useless ones got left in too. (laughs)

Koizumi: Yeah, like the crouching trip kick. (laughs) That was made to deal with short, small enemies, but we ended up not adding those kinds of enemies after all.

Getting a quick nap in.

—Were there any particularly difficult movements to create?

Koizumi: Movements that weren’t connected to any specific gameplay goal were difficult. It’s easy to design jumps, since Mario needs them to reach a certain ledge of a certain distance etc. But his sleeping animation, for instance, isn’t connected to anything else in the game, so it was harder to create an animation for. People told us “no one sleeps like this!”

Miyamoto: Mario knows how to sneak a nap in, they said. (laughs)

—I got used to the 3D stick’s analogue controls very quickly, and it felt completely natural to use in Mario 64, but I’m guessing it was actually very difficult to get right?

Miyamoto: It was a huge challenge, definitely. Now that they’ve had time to play it, people tell us it feels natural, but when we exhibited Mario 64 at the expo in November, we heard a lot of people say “I don’t know about this, the controls feel really wobbly and slippery…” We weren’t about to back down that easily, though. We dug in and pushed forward, knowing that this kind of response is to be expected if you’re trying to change the culture. And yet, while we were all telling ourselves “walking around leisurely can be fun too!”, I have to admit that internally, I was a little worried… normally games have a faster pace. (laughs)

Ultimately, though, we really did want to change the culture of gaming, and it was in that spirit that we made Mario 64. And that’s reflected in the controls, and how long it takes to accelerate.

—When I played that version in November, if I can speak honestly, my first impression was that I wanted the controls to be a little more responsive. But when I played the final version, I felt the responsiveness fit the game just right.

Miyamoto: We tightened the springs a bit on Mario, maybe? I don’t know. (laughs) In any event, though, Mario 64 required a lot of difficult physics calculations. This is an exaggeration to make my point, but when Mario is moving like a car, or moving like a human, or moving like an airplane… all those require different physics calculations. His different jumps all require different calculations too. It was very annoying.

—I remember you remarking that Wild Trax (Stunt Race FX) involved a lot of car-related mathematics too. Mario’s movement likewise has a proper sense of centrifugal force and “braking” friction… I noticed it when I was playing, and it reminded me of that conversation about Wild Trax.

Nishida: It’s not actually “real” physics, of course—we’ve fiddled with them a bit. Gravity, friction, resistance… these have properly set parameters, but if you rely on the actual laws of physics too much it doesn’t really make for a good game. The floatiness of Mario’s jump, for example, works great in a game, but a real airplane would never be able to fly at that speed.

Miyamoto: Yeah, it really was difficult! It was like we had to go back to middle school and learn basic physics concepts all over again—like what does it mean for a substance to be at rest? I never thought I’d have to study those things again at my age. (laughs) But that was also part of the fun.

—Whether we’re talking about Mario’s movements or the camera, I can see that working in 3D certainly has it’s own challenges. How about the aspect of game design?

Miyamoto: Well, I don’t think our game design process differed that much here compared with our 2D games. We spent about half our time and energy designing the basic system that we talked about. As for the courses and enemies, those actually came at the very end. They were done in a single burst of energy, just thrown together, almost. The hardware was very intelligent, so we had a lot of freedom in creating everything.

Yajima: The N64 hardware has something called a Z-Buffer, and thanks to that, we were able to design the terrain and visuals however we wanted. It gave us, as designers, a chance to play around with Mario in a diorama world, in a very free way. We could do a lot of experimenting—like, we’d make a ghost house course, and then drop Mario in there and see how it felt to move him around there. It was very fun.

—I know it’s a bit off-topic, but could you explain what the Z-buffer does?

Tanimoto: To explain it simply, when using it for 3D CG, it helps visually distinguish between foreground and background objects. If you have Mario and an enemy, and they move in front of or behind each other, the Z-buffer is what does the calculations and allows them to appear at the correct distances, with no weird flickering or overlapping.

Miyamoto: For example, without this hardware, if you had two characters A and B, you would only be able to calculate the position of A in a general way (not pixel-by-pixel). So when more complex or subtle positioning was involved, you’d get strange visual artifacts. To avoid that, the designers would have to make specific requests: don’t draw this kind of terrain, don’t make the camera do this, etc. There were a lot of restrictions. The N64 mostly doesn’t have those problems—”mostly”. Of course I don’t want people to think it was a breeze, either. (laughs)

Mario 64 clay diorama, taken from an unofficial strategy guide (image courtesy of Nintendo Life). The development team, interestingly, didn’t use any such blueprint maps, but created everything as they went.

—No, I think everyone can see how creating a huge 3D world like that would be a challenge! When you created the level maps, did you draw out models/blueprints beforehand?

Miyamoto: Actually, no, not at all. There would only be some concept art sketches, and brief notes/memos. For example, I’d talk with course director Yoichi Yamada about an idea for a level, then he’d make some quick sketches of it. Yamada isn’t an artist, but he draws weird stuff. (laughs) Then we’d look over those and talk more (“oh, there should be a snowman here!”), and those key elements of the level would be written down. Yamada and the other level designers then would refer back to those notes while designing the levels with our software development tools.

—Wow, you were able to create such complex levels with just a few notes like that?

Miyamoto: It was kind of like sculpting a diorama out of clay. First you make a very general shape. For example, with the King Bob-omb’s stage, for our initial, general design we’d have that river in the middle of the map, which you cross to reach the boss area, which is atop a big hill that you have to wind round and round as you ascend it. But say we put Mario in that map, and moving him around, we realize the river flows too fast and it sweeps him away (laughs), and that’s too hard for players, so we swap that river out for a desert valley, like Death Valley. So the form remains the same, but we gradually add more and more ideas, changing the map as we go.

Tezuka: Sometimes we’d be left with structure-like terrain without any purpose, and it’s like, ok—let’s just make it a little doghouse. (laughs)

Miyamoto: That was how we did it, basically adding the finishing details bit by bit. People kept writing notes with things they wanted to add, though… just seeing the notes, I had my doubts whether a lot of these ideas would work, but once they were actually created in-game, it was like, “Ok, I see how this works now.” (laughs)

—What were some of these doubts…?

Miyamoto: Well, for example, when someone had an idea for the race with Koopa, I was wondering how this was going to be made into a game. (laughs)

Tezuka: We had one early version where the course was more deterministic and restrictive of where you could go. You basically had to run up the mountain in a straight line. But it just didn’t feel like you were actually racing, so we changed it to the more freeform version where you have to dodge the balls as you ascend.

Miyamoto: Our original idea was that you’d race against a rabbit. But racing against a rabbit, which is so fast, would be too harsh and stressful. Besides, it would be more fun to make the player lose to a koopa. (laughs)

—Hugging the penguins, catching the rabbits… I loved all the little details you added that weren’t directly connected to the main game. It made me really happy to finally catch the rabbit—I walked around with him, tried releasing him… I played around with that for awhile.

Miyamoto: At first Mario was able to throw the rabbit too. (laughs) If we had another month, we could have added an animation where Mario tosses the rabbit by the ears… but we hit our time limit. I wanted to do it though.

Tezuka: I wanted to have more monkeys, too. In an earlier version of the game, we had them in more areas, and you could chase them around.

Miyamoto: If there were 3 of them together, they’d taunt Mario.

Tezuka: Yeah, and if Mario caught one of them, he could toss them off of a really high cliff. (laughs) I regret we weren’t able to do more with the monkeys.

—Were there other things you wanted to add, but couldn’t?

Koizumi: I wanted to have more tools, toys, and things for Mario to play around with: balls, cars, etc. I really think we could have worked a lot more of those in.

Miyamoto: We actually talked about including more frivolous elements like that: rolling a ball all the way up a mountain, or use it to plug up a gas flume, stuff like that. I think if you include a ball in a stage, then players would naturally want to try rolling it up a hill, right? You should tell us if that sounds boring, though. (laughs)

—No, I like doing stuff like that. With King Bob-omb, our editors tried to see how far we could roll him downhill. (laughs)

Miyamoto: That’s the kind of thing I want players to do! Pointless stuff like that. Truth be told, we did something with Mario 64 that we don’t usually do: we had children playtest it. We had a row of about 10 middle schoolers, and had them play around on the King Bob-omb’s stage for half a day, while we observed from behind.

My child was one of them, actually… but seeing him try dozens of times, over and over, to get up this unclimbable hill, as a parent I couldn’t help but think, “Geez, does this kid have any brains?” (laughs) Afterwards we asked the children what they thought of the game, and they said it was fun, and that they wanted to play it again.

Up to now, I think there’s been this image with games that if you can’t beat it, it’s not a fun or good game, right? That’s a philosophy we’ve stuck to at Nintendo, too, but I figured that if a game was this fun to play even if you weren’t getting anywhere, well, it must be alright. Until this game, I was very skeptical about something like this being fun.

—No, it really is a fun just wandering around doing nothing in particular. When I first played it, I spent awhile just running around the castle outside, swimming, jumping… it felt really good.

Miyamoto: That was our big gamble. We thought that half the people would just go straight into the castle, and half would hang out and explore outside, as you described. We made the game with that latter half of players in mind. I’m not saying that either way is “correct”, of course.

Bringing the baby penguin “Tuxie” back to its Mother. As noted by the team throughout this interview, non-traditional (for the time) gameplay elements like this were a huge part of Mario 64’s success.

—How about everyone else? Were there other things you wanted to do more of in Mario 64?

Yajima: I wanted to make terrain that could be changed—and for the game to remember the change. For example, if you destroy a block somewhere, and went to that area again, the block would still be destroyed. The N64 is capable of doing that, and I don’t think we’ve come close to exhausting it’s possibilities, to be honest.

Tanimoto: One of the things I programmed was the ripple effect when Mario hits a wall or jumps through a painting. This, too, is the kind of stuff that could only be done thanks to the N64’s processing speed. However, as we talked about earlier, Mario 64 was developed in parallel with the N64 hardware itself, so there’s a lot of untapped potential in the N64. Once we know it better, I think we can do much nicer graphics, and a lot more will be possible.

Miyamoto: That’s why I say this, and it’s no exaggeration, but we only used about 60% of the N64’s capabilities in this game. No, actually, if you look at everything, maybe only 40%.

Tezuka: There’s a lot of fun gameplay possibilities. In the middle of the development, we realized we could do so much more in terms of gameplay, and the staff were saying they wanted to hurry up and make a sequel!

Miyamoto: Let’s do it—but with Zelda this time. (laughs)

—Now that you mention it, Mario 64 did seem a bit Zelda-ish at times.

Miyamoto: For me, Mario and Zelda exist side by side. Their basic gameplay elements are the same, with the only difference being that one focuses on action, and the other on puzzle solving. They’re always developed at the same time, and lots of good ideas from Mario get used in Zelda, and vice-versa.

Actually, the castle system for Mario 64 was originally something I thought of for Zelda, and now that we’ve used it here, I’m wondering what we’re going to do for Zelda… (laughs) In any event, I’m really looking forward to developing the next N64 game. Uh oh, I better stop before I end up talking about what we’re making now. (laughs)

—We the players are also super excited about your next game!

Miyamoto: I’m glad to hear that, that’s good. We’re going to be releasing a lot of games that do things that have never been done before. We still have those doubt sometimes: “is this really going to be fun?”… but it’s precisely because it’s something weird that we want to try it! The N64 is that kind of hardware—it makes the strange possible. Please look forward to our next endeavor!