Street Fighter II – 1991 Developer Interview
Although the Street Fighter series is one of the better documented in terms of its origins and development, this 1991 interview with director Akira Nishitani is one of the earliest and most interesting records of the game’s creation available. Nishitani narrates the development from its roots in Final Fight to the last minute bugs and changes, noting interesting ideas that were abandoned along the way.
The Making of Street Fighter II
I’m going to back up for a moment here to talk about the Final Fight development. Final Fight was actually going to be Street Fighter II. I had been diligently working on the design plans for another game, but Capcom told me we needed to make something with less memory (the price of ROM had just gone up, they said), and they ordered me to make a typical belt-scroll fighting game. That was how I got started on Final Fight, and since our Street Fighter game had some name value to it already, I thought I’d use that title.
I wasn’t really too enthusiastic about this project, and so, as you might expect, when we exhibited it at the AM Show it was roundly criticized. That prototype was “Street Fighter 89.” Everyone knows it as Final Fight now, the name we eventually settled on.
Some time passed, and I was relaxing, not doing much after having finished Final Fight. At that time management approached me and talked with me about doing a sequel to the original Street Fighter.
I had all these ideas: a two-monitor cabinet in which players fight head-to-head? Or no, better yet, a series of networked cabinets, a 10 player battle royale! As for stages, how about a sandswept desert? Or the top of tower, where you fall to your death if you step out of the ring? Maybe an underwater stage where you have a limited air supply? Or a chain deathmatch where both fighters’ arms are bound! ………but what management wanted was a straightforward sequel to Street Fighter I. They didn’t want me to change the basic elements of the game. “Pfh, adults. All they care about is making a profit…. wait, aren’t I an adult too?”
For awhile I was pretty down about it, but eventually the contrarian in me reared up and I got more enthusiastic: “Alllright! If those are the limits, I’ll show them what a 100% Nishitani effort looks like!”
Well, now that I had decided to get down to it, I realized I had hardly played the first Street Fighter at all (maybe once or twice at the game center). Final Fight, though, I had played a lot of. Since I had so much Final Fight experience, I actually won the in-house Street Fighter I tournament at Capcom. It then occurred to me: why didn’t I play Street Fighter I at the game center? Because it wasn’t a very good game. “That’s it! I’m going to make a game so good, it will make the original Street Fighter look like a knockoff! I’m going to make a GOOD game that satisfies ME!” With that pledge in my heart, I began work on Street Fighter II.
From the very beginning I knew I wanted 8 player characters and 4 bosses. I thought that was a suitable lineup. The characters I first decided on were: Ryu and Ken, a wrestler (Zangief), a beast (Blanka), and a ninja (Balrog). For the others, I first decided on their nationalities, and used that to expand my image of them. The actual character names were decided on much later. During the development, here is what we called them:
Ryu = Ryu
Ken = Ken
E. Honda = E. Suzuki
Blanka = Beast
Guile = Soldier
Chun Li = Chinese Girl
Zangief = Vodka Gorbovsky
Dhalsim = Indo
M. Bison = Tyson
Balrog = Spanish Ninja
Sagat = Sagat
Vega = Washizaki
We decided to use a previously-unheard of 48MB memory capacity for Street Fighter II. The intention was to draw a clear line in the sand between previous games and this one. That meant, of course, that it would be expensive, and that we’d now have to sell a lot of copies. This put an enormous pressure on my shoulders, but there was no time to worry about such things at this late hour. Now that we had started there was only one direction: forward.
The character designers then started with rough sketches of each character. These became the basic designs we used. Next we decided on each character’s movements. How they walked, their damage animations, jumping animations, things like that. At this point we also needed to come up with every character’s moves and special techniques; with great care and great boldness we decided on their attacks. You had to take it seriously, because by this point about half the appeal of the game would be determined.
Finally the time came to actually put the characters together. We animated their footwork first, and put that up on the cabinet screen for everyone to see how they actually looked: the long-awaited moment. It was the first time seeing the entire team crowded around a monitor. This is the moment when most character designers are moved to tears, “Ah… my creation has finally appeared on-screen.”
At this point the character could now be moved around with the joystick. However, it’s just movement: there’s no hitboxes, and the height / speed / angle of their jump is very different from the final version. It was here that we decided to limit the number of Hadoukens on-screen to two; before that, we had no limits, and and it was really funny firing hadouken after hadouken. We played around with it, having three hadoukens of different speeds (weak, mid, strong punch) all on screen at once.
As we bustled about, time was passing, until that most dreaded of days finally arrived: the character deadline! And as we get closer to the deadline, things get rough for the character designers. Sleep itself has to be carefully rationed in order to finish their characters. What’s more, the amount of memory that seemed so huge in the beginning has now been almost entirely used up! “Um… there’s no more memory.” Well… then we have to cut the character’s entrance animations! (this was the point when we cut those out—only in Vega can you see a remnant of what we had done). Of course that was the easy part. A fierce scramble for memory took place after that: someone would remove a sprite from their character, and without a moment’s delay, another designer would come in to lay claim to the available memory.
There were numerous other trials, but humans somehow always find a way. And after this phase, things settled down for a time.
Balancing the Game
Here is where the planning and game design work really got busy. We needed to think up the overall game system as well as AI algorithms. Earlier fighting games didn’t have very good controls, and the computer AI was also pretty bad. Therefore, we took to improving the way the AI opponent moved and acted as our starting point for balancing the game.
I then made a crucial decision. “In Street Fighter II, your opponents are also player-controllable characters. It won’t feel right if the characters have a different moveset when the computer controls them. Alright, here we go! Time to develop a flexible AI algorithm.”
I wanted to create a system that was different than what had come before. Now that time has passed I can admit this though: it was a huge gamble, the gamble of a lifetime. My bosses and the programmers were all reassuring me, saying “it’s ok, don’t worry about it, it’ll work, it’ll work,” but the truth is I didn’t have much confidence in what we were doing, and to myself, I thought “I give this system a 50/50 chance of working out.” But sometimes as a game designer you need the courage to bluff.
Looking back, I can’t say it was a 100% success, but it did turn out well—thank goodness.
Then on a certain date, we had our long-awaited first location test. For a game designer, that night before is a mix of hopefulness of anxiety (although in truth there was so much work to do, you hardly had time to think). You didn’t sleep a wink that night. Will there be customers? What will the age-range be? How much income will it make? The great majority of the time, though, your expectations are betrayed. The unexpected will occur, and things won’t go as you planned.
Street Fighter II was not unusual in that regard. The difficulty setting turned out to be poorly tuned so that players were playing too long on a single credit. The controls were also too complicated, and players weren’t getting the hang of it. Most troubling of all was the storm of bugs. The screen would freeze… characters would suddenly fly from one area to another… the computer opponent would completely stop moving… and on and on. Each time there was a bug we’d have to investigate it and swap out the ROMs with a fix.
We had many more location tests, and I too spent a ton of time playing, determined to get the game balance right. But nothing in the world is ever so easy. I’d work out what I thought was the solution to a problem in my head, but once I sat down and got to work on it, numerous problems with my ideas would appear.
Consider for a moment: for each character, there are 10 possible opponents. When you think about the number of different combinations of attacks, you can see that it’s an astronomical number of things that need checking. Moreover, we’re not just talking about single punches and kicks, but complex combos. You fix one thing, and another doesn’t work… so many inconsistencies kept coming up.
In the end it came down to just doggedly playtesting the game over and over, but therein lies a pitfall. The more you play your own game, the more familiar you become with it, and strangely enough, you completely lose your perspective on what’s boring and what’s interesting, on what’s difficult and what’s easy (I think this is hard to understand unless you’ve been in that position as a creator of something).
When you have a stranger sit down and play your game for the first time, it’s often shocking (and worrying) how the things you expected have a completely different effect on that player. Up to the final deadline, I was still dragging over Capcom staff who hadn’t played the game yet, sitting them down and forcing them to play.
The one thing that we struggled with to the very bitter end was the AI for the computer-conrolled opponents. In the beginning my plan for the AI was to have it be able to do anything and everything, and I was going to to have each character be programmed with special characteristic behaviors. But by this point I had fallen into my own trap: we had given all the characters so many different attacks and moves that there was no way the computer AI could adequately respond to them all.
We had the most difficulties when the player used Dhalsim. We worked really hard on him, but in the end his abilities were just so different from the other characters that we couldn’t program the AI to handle everything. There ended up being a lot of cheap moves you could do with Dhalsim against the computer.
There we were, one month to go before the final deadline. When you’ve come this far, time limits what you can do. With a calm and collected judgment (actually, it was really all my own selfishness) I had to decide what would be cut due to time constraints. Here is a list of some of the plans we had that got abandoned (man, I really wanted to do these!)
* Add weak points depending on whether you hit someone in the head, body, or leg. If you hit someone there, they’d take more damage.
* Add other special weak points outside of those listed above (you can see remnants in Blanka’s Rolling Attack, or Balrog’s Flying Barcelona)1
* The computer would change its tactics depending on who it was fighting against (it does do this a little bit, but we wanted to do something more detailed, like the AI knowing how close to stand to each individual opponent, etc)
* Players would take more damage than normal when dizzied.
* If your opponent had lost almost all their life, and you knocked them out, there’d be a different special knockout animation (for example, if you did a jumpkick and your nearly-dead opponent also did a jumpkick, and you killed him with that jumpkick, there’d be a special knockout animation. One remnant of that can be seen in the KO animation when you die while guarding against a hadouken).
The well-known “uppercut –> shoryuken” combo was not something we originally planned. It came about as a chance side effect of some game settings we had adjusted. The first time I discovered it my pupils went wide with astonishment. I’ve been enchanted by it ever since then, but nowadays no player even lets themselves get hit by a shoryuken. Hey, somebody, let me do this move!
Now let me tell you about some of the mysterious special moves that we sadly had to cut out or abandon.
First, Chun Li. She actually had back flip attack.2 It was a move where she’d do a quick small jump and circle around behind the opponent to attack them. There may be some people who got to perform the move at one of the location tests, actually. I really wanted to leave it in, but it was too strong. We ran out of time to adjust it so we had to cut it.
Next is Blanka. The way he rebounds off the opponent when he connects with his Rolling Attack feels great. At first though, he would immediately return to the ground after it hit. At that point a very deadly technique could be performed, where you could go straight from the Rolling Attack into his Head Bite. Also, the way it controlled was different. Instead of holding back to charge it, you’d hold down to charge, then press foward.
Dhalsim’s Yoga Mummy and Drill Kick moves often trick players because they can be performed while he’s retreating. However, earlier in the development, when he did a back jump, those moves would come out in the reverse direction. You could use them to quickly escape, and it was funny when he did them near a wall… as for why we removed that, it was because the computer AI didn’t know how to handle it.
Then there’s Zangief. I know it’s a move that seems meaningless now, but his headbutt… it did have a purpose. It was supposed to be that he’d throw his opponent straight up into the air, and then headbutt him on the way down. But there were a lot of problems with it, so we cut that out. Only the headbutt itself remains as a remnant of the idea.
Finally, there’s Vega’s Deadly Throw. Being the Evil Emperor and all, we wanted to have him grab your fist and say something like “is that all you’ve got?” before he threw you. I really wanted that to be in the game. But it proved too difficult to actually show all that in-game, so we abandoned the idea.
One other thing I suddenly came up with at this late hour was the idea of a “fake” special move. It was something you could do without inputting the actual command for a special move. But the programmers were hesitant: “if you add that now, there’s no telling what bugs will come up.” But I made them add it anyway. The idea was that brand new players who couldn’t perform any of the special moves would learn of their existence that way.
We also had a feature where after you knocked out your opponent, there was a moment where if you kept continuously doing special moves, the screen would never advance… I remember challenging the other developers to see how many hadoukens we could do in a row to keep the match from ending (this was not a good time to be playing around).
11th Hour Bugs
With two weeks to go before the deadline, we were pulling all-nighters most every day, in an endless repetition of fine tuning –> bug check –> rom update –> print new rom –> find new bugs –> … We had to stop the production line numerous times, causing a lot of trouble for the people there. I’d like to take this space to offer my apologies to them.
Then, just three days before the deadline, I discovered something horrible. I had made a mistake with the subtitle “World Warrior”, mis-spelling it “World Warrier.” Now I can safely tell this story too, but we actually didn’t discover it until several months after all the sprite work had been done. Since the logo had already been created, I couldn’t just go in and change the letter at this point. “Maybe I can just force it to look like an ‘o’,” I thought. I tried layering various other sprites over it until finally it looked like an ‘o’. Phew.
This wasn’t the only horror that awaited us, however. Guile. He’s already a strong character to begin with, but he was actually even stronger before. In the finished version of Street Fighter II everyone knows, after performing a Flash Kick you land a little bit in front of where you started. But at this point in the development, after the kick you would instead land further back than you started. That really made it impossible to deal with, so I directed the programmers to change it. With one day before the deadline. I think they all thought I had lost my mind.
Eventually all the bug checking and fine tuning came to an end, and the skies started to clear up. Yes, I’ll never forget that date, February 14, 1991 (Valentine’s Day!), at 7AM in the morning, when the final version was at last completed. “It’s done, it’s over. I can’t believe it’s actually done…” A line of tears streamed down my face. I took everyone by the hand, exclaiming “We did it! We finally finished it!”
And so I entrusted the ROMs to a team member, and sent him directly to the production factory. I watched him walk away and slunk back under my desk. Muttering softly to myself “there’s no more bugs… no more bugs… no… there can’t be any more bugs”, I fell into a peaceful, if not eternal, sleep.
After the Release
The number one thing that concerned me after Street Fighter II’s completion was the invincible throw technique. We ran out of time to fix it, but I worried about it just as if it were an unexploded bomb, waiting to go off… When I went to the game center and saw people doing it, my heart throbbed with pain (it really did). So please, I beseech you, at least don’t do it in front of me! (probably an impossible request, I know)
The fighting instinct in human beings is amazing, isn’t it? I feel like that instinct has been unleashed with the release of Street Fighter II. Even though it’s only a game, I’ve witnessed some really intense gnashing of teeth when people lose a match. There were even some employees at Capcom literally writhing on the ground after losing several times in a row.
I haven’t seen many people playing against each other in Japan yet, but overseas it’s the opposite: hardly anyone plays alone. The battle counter is always at 99. Total strangers fight each other as if it were completely natural. I wonder if it’s because Westerners are a more warlike people? The way they fight in the game is very different too. They always want to attack instead of guard—it’s a sight to behold. I really loved this frank, open attitude, and I challenged many western players. It was extremely fun.
I want to be clear about one thing: I believe the appeal of Street Fighter II lies in playing against other human opponents. Lately a few game centers have started doing versus tournaments, but when strangers come up to challenge each other, they always give each other weird looks. So I want to say this now to all game center patrons: “Everyone, let’s do more versus matches! Come on guys, relax a little! Let’s just have fun playing these games!” (and at least challenge me if you see me!)
As a game designer, naturally what makes me happier than anything else is seeing the game I made in the game center, and seeing people there intently playing it. This isn’t my “public” stance either, it’s how I really feel inside. And if I see anything about Street Fighter II in a magazine I always read it. Man, you know, it feels just like my own child, it makes me really happy (geez, now I sound like an old man…). All the pain, overtime, late nights, and everything now seems like a happy memory that the game is released. So please give your love to Street Fighter II, now and forever…!
If you've enjoyed reading this interview and would like to be able to vote each month on what I translate, please consider supporting me on Patreon! I can't do it without your help!