The journey to create Street Fighter II wasn’t a straightforward path. Initially, I was immersed in designing a different game when Capcom redirected me to develop something less memory-intensive due to rising ROM prices. They requested a typical belt-scrolling fighting game, which led to the creation of Final Fight. Interestingly, Final Fight was initially conceived as Street Fighter II. Given the existing brand recognition of Street Fighter, I decided to use that title, even though my enthusiasm for this project was lukewarm. The prototype, exhibited as “Street Fighter 89” at the AM Show, was met with criticism, eventually being rebranded as Final Fight.
After completing Final Fight, I found myself in a period of downtime. Management then approached me about developing a true sequel to the original Street Fighter. My mind raced with ambitious ideas: two-monitor cabinets for head-to-head battles, networked cabinets for massive 10-player brawls, stages set in deserts, atop deadly towers, underwater arenas with limited air, and chain deathmatches. However, the direction from management was clear: a direct sequel to Street Fighter I, maintaining the core elements of the original game. “Adults,” I thought, “all they care about is profit…wait, aren’t I an adult too?”
Initially disheartened, my competitive spirit eventually ignited. “Alright,” I thought, “if these are the constraints, I’ll show them what a 100% Nishitani effort looks like!” To truly understand the task ahead, I realized I needed to familiarize myself with the original Street Fighter, a game I had barely played. My extensive experience with Final Fight, however, led to an unexpected victory in Capcom’s in-house Street Fighter I tournament. This experience was enlightening. I understood why I hadn’t frequented Street Fighter I in arcades – it simply wasn’t that compelling. “That’s it!” I declared, “I’m going to create a game so exceptional that it will make the original Street Fighter seem like a cheap imitation! I’m going to craft a good game that satisfies me!” Fueled by this determination, I commenced work on Street Fighter II.
Director / designer Akira Nishitani.
The very first concept map for Street Fighter II, presented in the fall of 1988, envisioned the game taking place on an abandoned island hosting a grand fighting tournament, attracting diverse fighting styles.
Crafting the Character Roster
From the outset, I envisioned a roster of eight playable characters and four boss characters. This felt like a balanced and compelling lineup. The initial characters I conceived were Ryu and Ken, a wrestler (Zangief), a beast-like fighter (Blanka), and a ninja (Balrog). For the remaining slots, I started by deciding on nationalities, using them as a foundation to build their identities. The actual character names came much later in development. During production, we used placeholder names:
- 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 made a bold decision to utilize an unprecedented 48MB of memory for Street Fighter II. This was intended as a clear declaration of intent, distinguishing our game from everything that preceded it. Naturally, this came with a significant cost, requiring substantial sales to justify the expense. This placed immense pressure on me, but there was no turning back. We had committed, and the only path was forward.
The character designers then began sketching rough concepts for each fighter. These sketches formed the basis of their visual design. Next, we focused on defining each character’s movements: their walk cycles, damage animations, jump animations, and so on. Crucially, this stage also involved developing each character’s unique moves and special techniques. With a blend of meticulousness and daring, we designed their attacks, understanding that this would define a significant portion of the game’s appeal.
Finally, the moment arrived to assemble the characters in-game. We animated their footwork first and displayed it on the arcade cabinet screen for the entire team to witness their creation come to life. It was a truly memorable moment, with everyone crowding around the monitor. For many character designers, seeing their creation move on screen for the first time is an emotional experience, a culmination of their work.
At this stage, the character could be moved with the joystick, but it was still rudimentary. Hitboxes were not yet defined, and jump heights, speeds, and angles were far from final. It was during this phase that we decided to limit the number of Hadoukens on screen to two. Initially, there was no limit, leading to humorous situations with players spamming Hadoukens endlessly. We experimented with allowing three Hadoukens of varying speeds (weak, medium, and strong punch) simultaneously on screen.
Early stage backgrounds. The text below says that environmental obstacles, like projectiles blown by the wind that you have to dodge, or water that you can splash at your opponent, were meant to influence and be part of the fighting itself. The Kanon statues in the left image would also sometimes throw things at you.
As development progressed, the dreaded character deadline loomed. As the deadline approached, the pressure intensified for the character designers. Sleep became a luxury, carefully rationed to ensure character completion. Furthermore, the initially vast memory capacity was rapidly dwindling. “Um… we’re out of memory.” The solution? Sacrifice character entrance animations. (Vega retains a remnant of these animations in the final game). This was just the beginning of a fierce scramble for memory. Designers would remove sprites from their characters, and almost instantly, another designer would claim the freed-up memory.
Despite numerous challenges, we persevered. After the character creation phase, a period of relative calm followed.
Balancing the Fight
The game balancing and AI development phase was intensely demanding. We needed to design the overall game system and create sophisticated AI algorithms. Previous fighting games often suffered from clunky controls and rudimentary AI opponents. Therefore, we prioritized improving the AI’s movement and behavior as a foundation for game balance.
I made a critical decision: “In Street Fighter II, the opponents will be controlled by the same character data as the player characters. It won’t feel right if the computer-controlled characters have a different moveset. Let’s develop a flexible AI algorithm.”
My aim was to create an AI system unlike anything seen before. Looking back, it was a significant gamble, perhaps the biggest of my career. While my bosses and programmers offered reassurance, internally, I estimated the system’s success at 50/50. However, as a game designer, sometimes you need to project confidence, even when uncertain.
In retrospect, while not flawless, the AI system turned out well – thankfully.
Concept art for Zangief, aka “Vodka Gorbavsky.” The text on the left says he “hates communism” and “became a street fighter for money, which is also why he fights in this tournament.” On the right, it says he was kicked out of official pro wrestling for ignoring the rules and permanently disabling all his opponents. He now makes a living in the underground wrestling circuit.
Then came the first location test, a nerve-wracking experience for any game designer. The night before was a mix of hope and anxiety, though the sheer volume of work often left little time for contemplation. Sleep was elusive. Would players be receptive? What demographics would be drawn to the game? What would the earnings be? More often than not, reality deviates from expectations.
Street Fighter II was no exception. The initial difficulty setting was too lenient, allowing players to play for extended periods on a single credit. The controls were also deemed too complex, hindering player adoption. Most alarmingly, a barrage of bugs emerged: screen freezes, characters teleporting across the stage, AI opponents freezing, and countless others. Each bug required investigation and ROM replacements to implement fixes.
We conducted numerous location tests, and I personally invested countless hours playing, determined to refine the game balance. However, achieving perfect balance is an elusive goal. Solutions conceived in theory often revealed new problems in practice.
Consider the complexity: each of the Original Street Fighter Characters had ten potential opponents. The sheer number of attack combinations to test was astronomical. Furthermore, we weren’t just dealing with single attacks but intricate combos. Fixing one issue often unearthed another – inconsistencies seemed endless.
Ultimately, rigorous playtesting became the solution. However, prolonged exposure to your own game can lead to a distorted perspective. Familiarity can blur the lines between engaging and tedious, challenging and trivial. (This is a creator’s dilemma difficult to grasp without experiencing it firsthand).
Concept art for Balrog (Vega). The text reads: “Top: Balrog was born from the concept of a ninja+matador. Left: Our original image for him as a crusader was abandoned in deference to Western religious sensibilities.”
Observing a newcomer play your game for the first time can be both startling and concerning, revealing how expectations can diverge from player experience. Right up to the final deadline, I would pull in Capcom staff unfamiliar with the game, sit them down, and have them play, seeking fresh perspectives.
The AI for computer-controlled opponents proved to be a persistent challenge. Initially, my ambition was to create an AI capable of handling every situation, programming distinct behaviors for each character. However, we had inadvertently created a trap: the sheer number of moves and attacks for each character overwhelmed the AI’s ability to respond effectively.
Dhalsim presented particular difficulties. Despite significant effort, his unique abilities proved so divergent from the other characters that programming comprehensive AI responses became impossible. This resulted in several exploitable “cheap” tactics players could use against the computer when playing as Dhalsim.
Ideas Left on the Cutting Room Floor
With only a month remaining until the final deadline, time constraints dictated necessary cuts. With a mixture of pragmatism and regret, I had to decide which features to abandon. Here are some of the planned features that didn’t make the final cut (features I genuinely wished we could have included!):
- Hit Zone Weak Points: Implementing weak points based on hit location – head, body, or legs – for increased damage.
- Special Weak Points: Beyond the basic hit zones, we considered additional character-specific weak points (remnants of this concept can be seen in Blanka’s Rolling Attack and Balrog’s Flying Barcelona). [^1]
- Adaptive AI Tactics: More sophisticated AI that would dynamically adjust its tactics based on the opponent character (while some basic adaptation exists, we aimed for detailed adjustments, such as AI managing distance based on the specific opponent).
- Increased Dizzy Damage: Dizzied players were intended to take significantly more damage than normal.
- Contextual Knockout Animations: Special knockout animations triggered by specific finishing moves when an opponent was near defeat (e.g., a jump kick KO against a nearly defeated opponent performing a jump kick would trigger a unique animation; a vestige of this remains in the KO animation when defeated by a Hadouken while guarding).
The now-iconic “uppercut -> Shoryuken” combo was not intentionally designed. It emerged as an unexpected consequence of adjustments to game settings. Discovering it for the first time was a moment of astonishment. It captivated me instantly, yet today, players rarely allow themselves to be hit by a Shoryuken. “Someone, please, let me land this move!”
The text reads: “Our first idea for an Indian character. He was named ‘Naradatta’ before we eventually settled on Dhalsim.” Naradatta may also refer to a character in Tezuka’s manga Buddha.
Let me share some of the more obscure special moves that were unfortunately cut or modified.
First, Chun Li was initially equipped with a backflip attack. [^2] This move involved a quick, small jump, circling behind the opponent for an attack. Some players might have encountered this move during location tests. I was keen to keep it, but it proved too powerful and we ran out of time to balance it effectively, leading to its removal.
Regarding Blanka, the satisfying rebound effect after his Rolling Attack connected was a later addition. Initially, he would immediately return to the ground after the hit. This enabled a highly damaging technique: transitioning directly from the Rolling Attack into his Head Bite. The charging mechanic was also different; instead of holding back, players would charge by holding down and then pressing forward.
Dhalsim’s Yoga Mummy and Drill Kick moves are known for their deceptive nature, usable while retreating. Early in development, however, these moves would execute in the reverse direction when performed during a back jump. This allowed for rapid escapes, sometimes comically near walls. We removed this functionality because the computer AI struggled to handle it.
Zangief’s headbutt, a move that might seem pointless now, was originally intended to have a purpose. It was designed to launch opponents vertically into the air, followed by a headbutt on their descent. However, due to various issues, this concept was abandoned, leaving only the headbutt itself as a vestige of the original idea.
Finally, Vega’s Deadly Throw was envisioned with more flair. As the grand villain, we wanted him to catch the opponent’s fist, deliver a taunt like “is that all you’ve got?”, before executing the throw. Implementing this in-game proved too complex within the time constraints, so the idea was scrapped.
One last-minute idea I conceived was a “fake” special move – a move that could be performed without the actual special move input. The programmers were hesitant, fearing potential bugs. However, I insisted, believing it would introduce new players, unfamiliar with special move inputs, to their existence.
We also had a glitch where continuous special move inputs after a knockout would prevent the screen from advancing. I recall challenging other developers to see how many Hadoukens we could consecutively execute to prolong the match (perhaps not the most productive use of time).
Concept art for Chun Li. The text reads, “Our image for Chun Li changed from the somewhat coquetteish ‘Chinese Girl’ version (left) to the strong, gallant Chun Li (right).” Note the Interpol outfit.
Last Minute Bugs and Valentine’s Day Victory
With two weeks remaining before the deadline, all-nighters became routine, a cycle of fine-tuning, bug checks, ROM updates, printing new ROMs, discovering new bugs, and repeating. We had to halt the production line multiple times, causing significant disruption. I sincerely apologize to those affected.
Then, just three days before the deadline, a horrifying discovery: I had misspelled “World Warrior” as “World Warrier” in the subtitle. Now, years later, I can share this story. We didn’t catch it until months after all the sprite work was completed. With the logo already finalized, a direct correction was impossible. “Maybe I can force it to look like an ‘o’,” I thought. Through layering various sprites, I managed to visually transform the ‘r’ into an ‘o’. Relief washed over me.
However, another crisis awaited. Guile. Already a powerful character, he was even more dominant in this late development stage. In the final Street Fighter II, a Flash Kick results in Guile landing slightly ahead of his starting position. But at this point, he would land further back. This made him nearly unstoppable. With just one day before the deadline, I instructed the programmers to change it. I suspect they thought I had lost my sanity.
Eventually, the bug fixing and fine-tuning concluded, and the skies began to clear. Yes, I will never forget that date: February 14, 1991 (Valentine’s Day!), at 7 AM, the final version was complete. “It’s done, it’s over. I can’t believe it’s actually done…” Tears streamed down my face. I grabbed everyone’s hands, exclaiming, “We did it! We finally finished it!”
I entrusted the ROMs to a team member, sending him directly to the production factory. Watching him depart, I slumped back under my desk, muttering, “No more bugs… no more bugs… no… there can’t be any more bugs,” and drifted into a peaceful, if not eternal, sleep.
Certainly not the most egregious Engrish to be found in games…
Life After Release
My primary concern after Street Fighter II’s release was the invincible throw technique. We ran out of time to fix it, and it felt like an unexploded bomb. Seeing players exploit it in arcades caused genuine physical pain. So, I implore you, please, at least refrain from using it in my presence! (Perhaps an unrealistic request, I know).
The fighting instinct in humans is remarkable, isn’t it? Street Fighter II seems to have unleashed this instinct. Even though it’s just a game, I’ve witnessed intense displays of frustration when players lose. Some Capcom employees literally writhed on the floor after consecutive defeats.
In Japan, I haven’t seen much player-versus-player action yet, but overseas, it’s the opposite. Hardly anyone plays solo. The battle counter is always maxed out at 99. Complete strangers challenge each other as if it’s the most natural thing in the world. Perhaps Westerners are simply more combative? Their in-game fighting style is also distinct – always prioritizing attack over defense, a captivating sight. I genuinely enjoyed this direct, open approach and challenged many Western players myself. It was incredibly fun.
Let me be clear: I believe the true appeal of Street Fighter II lies in player-versus-player competition. Recently, some arcades have started hosting versus tournaments, but when strangers approach each other for matches, they often exchange awkward glances. So, to all arcade patrons: “Everyone, let’s engage in more versus matches! Relax a little! Let’s just enjoy these games!” (And feel free to challenge me if you see me!).
As a game designer, nothing brings me greater joy than seeing my game in arcades and witnessing people engrossed in playing it. This isn’t just a public statement; it’s my sincere feeling. And whenever I see anything about Street Fighter II in magazines, I always read it. It feels like my own child; it fills me with happiness (I sound like an old man…). All the hardships, overtime, late nights – they now seem like cherished memories now that the game is released. So, please, show Street Fighter II your love, now and forever!
[^1]: [1] The original text mentions Balrog as the ninja, but this is likely a localization confusion. Balrog in the Japanese version is M. Bison in the English version. Vega in the Japanese version is Balrog in the English version. This refers to Vega’s (Balrog in English) Flying Barcelona Attack.
[^2]: [2] Chun Li’s backflip attack may have been repurposed into her Spinning Bird Kick, which shares some visual similarities.
