happa

The honors of making a true 3D adaptation of Final Fight and belt scroller beat ‘em ups are often bestowed on God Hand or the ill-begotten Final Fight Streetwise, but few know of the attempts that came before it. In 1996, Dynamite Deka / Die Hard Arcade came out as the first ever 3D beat ‘em up. It featured 3D polygonal graphics, but in terms of gameplay it was a very literal translation of belt scrollers to 3D–down to having your facing directions be limited to only left and right. In 1997 it would receive a sequel, called Dynamite Deka 2 / Dynamite Cop. It allowed for eight-way movement and attacking, but the levels themselves were by and large flat squares. That same year, the Tomb Raider developers released Fighting Force on the PS1. This one featured a third-person camera with more complex 3D environments, but it fell flat with its enemy and behavior design, which really lacked the means to pressure you effectively in 3D space.

But in 1998, the madmen over at SEGA would leverage their expertise in 3D gaming and the sheer absurd power of the Sega Model 3 to the at-the-time waning genre of beat ‘em ups. Stuck on one of the harder-to-emulate arcade platforms that rarely ever appeared in Western arcades, and with an obscure and forgotten Xbox OG remake/sequel as the only home system release¹, it was forgotten by the public for years. Until finally, a port of the arcade version in the newest Yakuza game reignited some interest. This game was called… SpikeOut: Digital Battle Online.

i. first sidestep towards wars

SpikeOut: Digital Battle Online (and its Final Edition update) is the true complete package: lethal enemy crowds that demand respect and stick to you like glue, a combat system that emphasizes spacing and whiff punishment, crowd control that relies on both aggressive i-frame deployment and manipulating crowds via positioning, a juggling system that allows for damage optimization and flexibility in picking up enemies without being either too easy to pull off or overpowered as a strategy, a strict timer refilled by K.O.ing enemies/damaging bosses that puts a hard cap to how much cheese you can afford to spend time on, and get this, bosses with supporting minions that are not reliant on hidden hard counters to your movesets or some other gimmick that disregards the game's fundamentals. And all of it wrapped in those pleasing vibrant Californian early 3D SEGA aesthetics, with some neat background choonz to top it off. But what makes SpikeOut interesting isn’t just that it’s a good beat ‘em up, but how committed it is to making full use of 3D space.

Translating belt/side scrolling beat ‘em ups (subgenres where usually a character can only face either left or right) to 3D (where characters can face almost any direction) requires a strong rethinking of genre fundamentals and its notions of offense and defense for situations to not become uncontrollable for the player. For example, in what ways should enemy groups attack the player in 3D? What role could complex 3D environments even play in combat when most of the fighting takes place at point-blank range? What kind of toolset does the player need to control 3D space? What kind of camera should we use to frame 3D action?

Most of the first 3D forays involved radical changes to its fundamentals. Musou games, the God of War games, and SpikeOut's fantasy-themed sequel SlashOut dramatically expanded the width and range of your basic strikes to more effectively cover 3D space. Games like DMC1 and Bayonetta would give you permanent access to ranged attacks and more i-frames on demand to deal with enemies coming from any direction, while games like Oni and Ninja Gaiden gave you powerful block buttons or mobility options. Fighting Force and the Dynamite Deka games, while overall staying close to beat 'em up fundamentals, disappointingly opted to limit enemy aggression and the maximum number of enemies on screen to about four–way below what even Nekketsu Kōha Kunio-kun and Final Fight could render in the late 80s. Although to be fair, not having 3D-capable hardware as HUMONGOUSLY powerful at the time as the Model 3 must have also factored into that decision.

SpikeOut sticks out from all these precisely because of how tightly it adheres to beat ‘em up fundamentals. Your main attacks have relatively pitiful range, enemies always move faster than you, you are always outnumbered, enemy attacks are often not reactable, disengaging from crowds is not free, the main method of defense is to simply move out of range, and i-frames are earned rather than given. Still, a basic setup of jabs, jump kick pokes, and grabs that grant you invincibility isn't going to be as effective in 3D space when you have to worry about attacks from all 360 degrees instead of to your immediate left and right.

One of the most important details to consider is that in 3D, enemies can't overlap with each other or occupy the same space. In 2D one enemy would simply visually overlap the other, but if you did that in 3D they'd clip through each other, which would look grizzly. Enemies being unable to overlap also means that enemies are fundamentally going to be spaced and spread out from each other more. The implicit 2D victory condition of rounding up enemies into one big pile and looping them to death is simply not possible by default anymore. Jab strings and jab infinites in general are going to be much weaker if the space covered by your attacks remains relatively the same, so at best you can only keep one enemy stunlocked at a time. And if enemy behavior is such that if you miss an attack or hit an enemy close to another, it's basically guaranteed you are going to get punished with a counterattack.

To this end all player characters in SpikeOut received some necessary new moves to control these more spaced-out crowds more effectively. You get a sweep meant for hitting horizontally-lined up enemies, you get a homing jump slam that serves as an (unsafe) guaranteed approach tool, and more importantly: your charge moves. Depending on how long you hold the charge button, you can perform: a C2 launcher that launches every single enemy type in the game without exception; a C3 stun blow that leaves all hit enemies in an extended stun state, and a powerful C4 bowling strike that with the right timing, space, and firing angle can knock over entire crowds like bowling pins. You can imagine that a move like C3 would have been rather busted if enemies could occupy the same space and be hit all at once. Likewise, a move like C4 that knocks down all enemies in front of you would have also been rather busted if all enemies had to align with you on the Z-axis to be able to hit you at all. But because this is 3D, you need to actively manipulate crowds through movement to even get them to line up at all.

Being able to line up enemies is only made possible because SpikeOut actually nerfs general enemy behavior from belt scrollers in one key aspect: most enemies aren't going to try and circle around you. Some bosses and enemies do have this ability, but on most regular enemies this was left disabled, and for good reason. Think back to Resident Evil 4, and how for all the beat 'em up traditions Resident Evil 4 translated to a third-person shooter format, enemy circling is one it also left out! Could you imagine trying to keep track of circling enemy crowds using the original PS2/GC controls? Now, in 2D belt scrollers having most enemies circle around you was necessary to prevent you from always being able to backpedal into safer territory. Since every character could only attack directly left/right and therefore always keep one side covered, one way enemies could then break your space control was to move towards your uncovered flank. Otherwise you could hang back and have enemies be forced to walk into your jabs, which is also why actual walls to cover your back with aren’t always present in belt scrollers. But when we're dealing with 360 degrees of freedom and your standard attacks only covering about 5 to 10 degrees, having enemies always circle around you and attack you from multiple angles would quickly turn into an uncontrollable nightmare. Imagine, you could be throwing an enemy while the other five have already aligned in a kung-fu circle around you, the ones outside the range of your throw ready to clip you right after your throw invincibility ends. Aggressive i-frame deployment would become a total liability under such conditions, leaving a backpedal-and-poke playstyle as the only viable option (if you’re out of limited space-clearing specials). The trio boss fights in SpikeOut are good examples of what dealing with a group of fast enemies who constantly circle around you would look like, and it's not pretty.

That covers how the enemy approaches the player, but how does being in 3D affect the player approaching the enemy? Consider how you’d approach enemies in a belt scroller: if all characters can only face two directions, and being in range and aligned on the Z-axis with an enemy triggers an unreactable attack, then one safe approach is to use long-range pokes such as forwards jump kicks (which run the risk of being intercepted with an uppercut or having their recovery punished), or alternatively, approaching an enemy diagonally. You’d move off-axis towards an enemy where their attacks can’t reach you, but meanwhile other enemies would move towards or away from you. This jockeying for positions and angles is then what drives the “neutral game” of a belt scroller. But how is this going to work in 3D when all enemies are always facing you by default? There is no way to move “off-axis” now. Moving in a straight line into an enemy’s face was guaranteed to get you slapped in a belt scroller, but now they’re always facing you. With no way to safely approach beyond using a (punishable) poke, it would seem like your best bet is to wait for enemies to move into your attack range.

SpikeOut’s solution against that is rather subtle: there is now a greater emphasis on whiff punishing enemies. In belt scrollers most enemy attacks would start and recover too fast to be reliably whiff punished, so in SpikeOut the average recovery and duration of enemy attacks has been significantly extended. Instead of moving off-axis for a safe approach, you create an opening via baiting an enemy into attacking by sidestepping through their attack range and see their attack whiff. You can keep circling around most lone enemies and basically never get hit (that is, unless they can bust out grabs or spin attacks), since most enemy attacks don’t have massive hitboxes or tracking. Of course, what prevents you from being able to do this whenever you want is that in most cases there are other enemies standing right beside one another. Circle around one enemy and you will find yourself in the attack range of another enemy, circle around him to find yet another, and keep doing this long enough and you’ll eventually eat a homing kick to the face. Or, your path will simply get blocked by a wall. Here the aforementioned choice to have enemies intentionally not circle around you but instead clump up together synergizes wonderfully: if enemies spread out from each other around you, then they couldn’t as effectively cover each others’ blind spots and prevent you from whiff punishing enemies uncontested! It’s then up to the player to create an opening in the crowd before they can really start focusing on whittling down individual enemies.

Another curious change is how SpikeOut handles variable enemy wakeup timing–or rather, how it doesn’t handle it at all. In belt scrollers it was generally a good thing to have enemies wake up at different times in order to prevent you from easily rounding up enemies into a pile and looping them to death.² Now consider how this is supposed to work in 3D or what the point of it even would be, when enemies can no longer overlap the same space. Enemies waking up at the same time in SpikeOut ironically ends up making things more difficult for you. Should you knock down multiple enemies with a sweep and try to meaty one of them, then depending on how you positioned yourself, the other enemies getting up at the same time can quickly jab you during the recovery of your meaty attack. Knocking over enemies at the same time ends up being something you want to avoid, and juggling enemies ends up being something you do not just for the damage but also to ensure multiple enemies recover at different times.³

Then there's the particular matter of SpikeOut's controls. One thing will immediately stand out: turning your character about and around ain't instant. Which if you think about it, is the case for a lot of 3D action games where character orientation isn't controlled separately from movement (as you would with a mouse or right joystick). Ninja Gaiden, DMC, Bayonetta, take your pick. If you try to turn 90 or 180 degrees there is usually a maximum turning rate or a special turnaround animation, and SpikeOut is no different. One reason being that a 3D character instantly flipping or turning around looks off when all other animations in the game use smooth interpolation, especially when the camera is closer to the character. It'd be rather disorienting if you were trying to approach an enemy from behind only for them to almost instantly turn around. For the sake of fidelity we can then implement a maximum turning rate, but this will come at the expense of the pixel-perfect movement precision and responsiveness you’d normally get in a 2D belt scroller. 3D games have found a lot of ways to compensate for this: dodging/blocking abilities with wide coverage, larger hitboxes and/or magnetism on player attacks, reduced aggression on enemies, or more commonly: strafing movement via a hard lock-on button. SpikeOut does apply some magnetism on your attacks (on the horizontal axis only), but your main lifeline for precise control in SpikeOut is called Shift.

Shifting is basically Ocarina of Time's Z-targeting, except SpikeOut came out 5 months prior (although Nintendo demonstrated OoT’s Z-targeting at E3 a year prior, so who knows what came first). Not only does it let you move more precisely and circle about enemies while keeping your character's orientation fixed on the crowd, and not only does it also let you walk in perfect circles and navigate the not-so-rectangular level layouts using SpikeOut's 8-way digital joystick, but it also has a higher base movement speed than normal walking (albeit lower than running). In fact, if you are waking up with an enemy behind you, it is usually faster to avoid their meaty by shifting forwards away from them than it is to try doing a backsweep or turning around and jab, as you can't move while doing a quick turn. Another thing about shifting is that you’re locked out of grabbing enemies while doing it, which, in a 3D game where you grab enemies by walking into them, is really really useful to have. Accidental grabs can already happen quite often in belt/side scrollers, but at the very least you can throw said enemy to clear some space. Having accidental grabs happen in 3D where your throws don't cover as much space would be just grueling, so being able to toggle walk-in grabs when trying to move out of a crowd is a godsend.

But the most pressing justification of shifting/lock-on existing in SpikeOut is enabling you to more precisely aim the position and angle for your C4 attacks. With your orientation already fixed, you only need to move left/right to adjust your angle and forwards/backwards to adjust your 'firing' position. In a lock-on-less control setup where you’d instantly turn towards your movement direction, you'd have to reorient (and therefore slightly move) yourself towards the enemy each time you adjust the angle or position of your C4, effectively creating some inaccuracy. In games where the player can only face 2/4/8 angles this inaccuracy isn't a big deal, but with 360 angles, especially with the hitboxes on C4-launched enemies being as narrow as they are, it kind of is. You could then increase the width on the enemies launched by C4 to compensate for this inaccuracy, but that is only going to create different balance issues and possibly make it too strong. As they already are, C4 attacks strike a good balance between their power/effectiveness and the execution required to make the most of them.

ii. touch fuzzy, get dizzy

Now, the first obvious thing anyone will try to do in SpikeOut is to spam the C3 and C4 charge moves to oblivion. This might carry you for the first few areas, after which you are forced to come to grips with two facts about SpikeOut's universal enemy behavior:

1. In a neutral state, most enemies can backstep away on reaction if you try to hit them with a charge move, and only charge moves specifically (and also jump throws in Final Edition). Landing charge moves is not free.
2. If you move away from enemies too much they will turn into metaphorical homing missiles. They will outrun you at a speed slightly faster than your fastest run speed, and then hit you with a homing kick/slide/tackle that needs to be sidestepped or interrupted with an attack at the right time. On Hardest Difficulty, absentmindedly moving backwards for a second is enough for enemies right in your face to spontaneously bust out a homing kick without warning. Disengaging to find time and space for charging your charge moves is not free.

At first glance these come off like static interactions which the player can only (not) react to in a few clearly predetermined ways, but in practice their implementation is so tightly woven into time and space–and therefore everything else tied into them–that more than anything the options and outcomes for any situation or interaction are fuzzy. An enemy can decide to backstep away from your charge move 80% of the time, but what if they are blocked by the environment? What if they are bodyblocked by another enemy with a lower backstep chance or an enemy that wasn't in a neutral state? What if you hit them with a charge move right after they wake up from a knockdown? What if you just so happen to release your charge attack right as they started an attack of their own? What if you bowl them over by launching into them an enemy that rarely ever backsteps anyways? What if you launch them after first knocking them down with a homing jump slam that they can't backstep from, and is the situation such that that the jump slam itself is safe?

Likewise, every enemy type being hardcoded to eventually chase you if you move backwards can be used to your advantage. It can be used to bait any enemy type into doing a single predictable action, but the question remains if you got the time and space to do so. You need time and space to sidestep, and doing this for more than a second against a crowd of enemies will result in multiple enemies trying to hit you with homing kicks from multiple angles and timings. On top of that, when enemies start chasing you is (as far as I could tell) triggered semi-randomly, so you can’t even predict with full certainty either the when or where of enemies entering a chase state. Even more interesting is that you can force enemies to cancel their chase by standing still or moving forwards; during their chase recovery animation they are completely vulnerable and can’t backstep from charge moves. This requires good timing and spacing to pull off and take advantage of, but the exact threshold for when they will cancel their chase is also vague.

That’s the thing with SpikeOut. Not only is the combat such a sensitive clockwork machine where even the slightest change in your actions creates unforeseen consequences down the line, but you also have the actual RNG adding further chaos into the mix. Nor do you have any overcentralizing abilities like an omni-directional dodge or a parry that could reliably mitigate all this chaos. Whatever’s the best approach in any given situation ends up being fuzzy, and this fuzziness extends to the grand majority of interactions. For better or worse, nothing is guaranteed to succeed, nothing is free, everything has several edge cases and everything depends.

Consider how time complicates things. An example is the literal timer itself, which is a bit special. It counts down rather quickly by arcade standards, such that if you spend too much time slowly cheesing enemies by baiting them into doing homing kicks, you'll likely game over by timeout. However, it is refilled by K.O.'ing 8 regular enemies or a boss enemy, so this constant time pressure puts a hard limit to (but doesn't entirely forbid) the cheesing you can do, while also pushing you to take risks you otherwise wouldn't have if you are close to running out. A smaller example of time inducing fuzziness is in the timing of inputting strings. While you can mash out a basic BBBBBB string, more damaging strings (like BBBCCC) and keeping enemies stunlocked with jab strings (like BB-BB-BB or BBBCC-BBBCC) requires a very strict rhythm where messing it up slightly creates enough of a time window for the enemy you were stunlocking to interrupt you. While it's theoretically possible to have a perfect rhythm, in practice there's always something that will eventually mess up your rhythm and leave you scrambling trying to recover from it--something much less likely to happen if the input timings for strings were significantly more lenient.

But perhaps one of the strongest examples of time as fuzziness is the way the charge gauge itself works. A beat 'em up developer's first instinct would have been to have moves as powerful as your charge moves be executed via motion inputs or lock them behind resources/cooldowns, meaning you can make it come out as fast as you can input it or can spend the resources for it, such that the aspect of time spent in landing the move itself plays a much smaller role. But by tying different charge moves to different charge time windows they are more tightly woven to Time itself, and therefore everything else connected to it.⁴ So it's up to you to judge whether you have the time or space to charge a C4, because your enemies aren't going to make it cheap. An enemy deciding to pressure you while charging can force you to decide whether you want to go for a pre-emptive C2/C3 or risk going for the C4. Nor can you hold a charge forever, as you can only maintain a maximum charge for 2 seconds before the gauge resets. Nor can you sidestep the time pressure by charging while running away, because charging while running (and jumping) is explicitly disabled, and enemies will chase your ass if you do turn tail.

Of course, then there is also realizing fuzziness via space. For example: the arc on your homing jump slam is calculated at the start of the move, meaning that if the enemy suddenly moves or already was moving after you initiate it, the jump slam will likely whiff. To manipulate them into not moving you need to keep proper distance and not back away, and then it depends if the enemies give you the time or space to do that! Then take as another example SpikeOut's "strike grabs". In Final Fight 1, if you held UP during the last hit of your jab combo you'd consistently back-throw whatever enemy would have been hit even if they were outside grab range. But in SpikeOut? It depends. Here there are no 'true' strike grabs, and you cannot grab an enemy that is in an attacking state. Whether you can safely grab an enemy after a double jab depends on how close you were to the enemy afterwards, so it's up to you to judge whether you need to do another double jab to close the distance or change plans entirely.

iii. dimensional

And what else could contribute more to the fuzziness than adding an entire new dimension? Stages in SpikeOut aren't all flat planes: they've got complex layouts, they've got undulation, they've got multiple floors, they've got moving floors, they've got level geometry in the middle of a stage, they've got throwable physics props that can collide with each other and impede you, they've got escalators, they've got an area with a giant Sonic the Hedgehog statue in the middle, and they've got--god forbid–corners. In addition, your elevation relative to enemies now matters because your attacks don't track or extend downwards on the vertical axis, which counterintuitively makes having the high ground on enemies actually a disadvantage and fighting on slopes or spaces with undulation tricky. Another consequence of that is there being minor high crush / low crush shenanigans a la Tekken where because the hitboxes are tied so accurately to character models, certain player and enemy attacks can duck under high attacks. Then consider that health/special attack items also have physics applied and can bounce down a flight of stairs. Juggling enemies is now made a bit more complicated with the existence of walls and wallsplat states. Just ask any Tekken fan about the consistency of off-axis interactions near walls. There is even a merry-go-round area at one point where the floor is split in three rings turning at different speeds!

And... for some reason launching enemies into walls sometimes nets you a wallslump state with no possible follow ups instead of the usual wallsplat state from which you can continue juggling. And if you throw a weapon alongside a wall with your throwing arm too close to the wall, your weapon will collide with the wall and just drop on the ground. And the collision detection will sometimes shit itself if too many bowled over enemies are colliding with each other in a tight space. And also enemies easily get stuck everywhere because their pathfinding can't move around obstacles. The third dimension may make interactions more complicated, but It's not all flowers and sunshine.

You rarely see these kinds of more involved environments in melee games nowadays. The trend in modern melee games has been to abandon complex environments in favor of clear open flat arenas. These are much less likely to create inconsistency or bugs for player-enemy interactions and--perhaps more importantly--don't obscure the camera. Fights in older DMC games tended to take place in more distinct areas, and by DMC5 the majority take place on indistinct flat open spaces. Tekken and Virtua Fighter both once experimented with stage undulation and more complex environments (coincidentally, SpikeOut's lead programmer was also the one responsible for programming the undulation in VF3's stages) before reverting back to flat arenas for all future games. The inconsistency and unpredictable edge cases it wrought proved undesirable for high-level competitive play, whereas SpikeOut as a singleplayer/co-op romp seems to embrace the chaos. Certainly simpler layouts are superior from a quality-of-life standpoint both to develop and to experience, but there's also certainly something about levels feeling like actual places instead of flat killboxes. And as mentioned before, it allows for more varied layouts to create more varied encounter scenarios. So fighting crowds of enemies in a small tight square gives you barely any room to breathe, long narrow spaces on the other hand can be used to your benefit to group up enemies and send them all bowling over with a C4, and areas with height differences introduce more chaos. Had SpikeOut been so inclined, it could've even used the new possibilities in stage layouts to introduce platforming to the mix. ⁵

SpikeOut's strong grasp on beat 'em up fundamentals and the strength of its own system stands out even more when you notice there is barely any "overt" variety or gimmicks or paradigm shifts in enemy behavior, yet the game rarely ever feels stale even on the umpteenth replay. Although every 2-3 areas the previous roster of enemies gets thrown out and a new unique set of enemies is introduced, the behavior of each new enemy remains largely universal. Enemies might have particular attacks and stats and traits, but how you individually deal with them and how they deal with you remains largely the same, except for the many boss enemies who instead have their own universal set of rules. It’s why you can’t really pick out a single good or bad enemy type in SpikeOut–most of their behaviors are basically the same. Compared to something like Final Fight 1 or Streets of Rage, you don't have enemies that specialize in doing a particular thing, such as guarding or grabbing you or jumping behind you or ramming you or throwing shit at you or just flying around the place, and the game isn't like a Treasure game where there's a paradigm shift every two minutes⁶. There are some enemies in SpikeOut that are specialists, such as enemies wielding weapons that can't be knocked out of their hands or the flamethrower guys that can sweep the area in front of them, but those are the exception rather than the rule. It's more accurate to say that almost all enemies draw from the same universal behavior, but have their dials slightly tweaked this and that way.

And yet, the fuzziness of SpikeOut's gameplay makes it susceptible enough to the butterfly effect that even minor tweaks in enemy behavior and traits combined with slight offsets in spawn patterns, level geometry and other factors are enough to cause unique scenarios to spiral out by themselves. One thing leads to another which spirals out in yet another and you’re back in unknown territory where memorized routes can’t (completely) save you. For a game made to be replayed infinitely, this is ideally the way to go. In another context where player-enemy interactions were made more consistent/explicit/controlled and resistant to outside interferences (so things like attack magnetism, lenient super armor/invincibility for the player and/or the enemy), or if we’re dealing with 2D space instead of 3D, it would be more unlikely that such minor fluctuations could spiral out into something major, and it would be more unlikely that identical interactions with identical enemies could remain interesting across multiple replays. It certainly wouldn’t have hurt if SpikeOut added more specialized or gimmicky enemies, but it’s nonetheless impressive that SpikeOut never gets boring even without them.

Now, all this fuzziness is certainly nice for creating uncertainty in replays and drawing on your skills to read the situation and improvise accordingly, but it can also make survival feel inconsistent simply because 3D space adds more room for error (try imagining having to dodge a bullet hell pattern in 3D). There will be unforeseen edge cases biting you in the ass, or enemies sneaking in jabs and potshots from more angles than you can reliably cover. SpikeOut theoretically gives you all the tools you need to succeed in 3D, but being in 3D also gives you more room to mess up how you use those tools, like getting the angle on a C4 or sweep slightly wrong. Considering clearing arcade games usually involves being able to play it consistently without taking too much damage, then avoiding damage being too chaotic to do consistently will quickly result in a frustrating experience. You could theoretically clear SpikeOut without taking a single hit, but with the timer being as it is that would only be possible with a TAS. At the same time, reducing the fuzziness of the gameplay to be fundamentally more consistent would come at the expense of its replay value, so some kind of balance needs to be struck.

Here SpikeOut manages to strike a good balance by biasing certain outcomes in the player’s favor through ways that would be overkill for most 2D belt scrollers. For example, your C4 and homing jump slam–both being your two moves with the most wind-up– have light super armor, meaning they won’t get interrupted by light jabs (but will be by heavier attacks). You are also briefly invincible while waking up to prevent you from being stunlocked or having to bust out a special, which is common for platformers, but actually quite rare to see in beat ‘em ups. And more importantly, you get a big health refill after clearing an area with a boss fight (every 1-2 areas or so). This health bonus is determined depending on how many enemies you K.O’d, and while playing solo this usually means getting 60%-90% of your health back. Crucially, SpikeOut opts to be more forgiving when you mess up instead of preventing you from being able to mess up at all, while still being only lightly forgiving in a way where you can’t intentionally try to rely on it.

epilogue

SpikeOut is really interesting to analyze, not only because it is a rock-solid game that successfully translates genre fundamentals to 3D, but also because it shows us this alternate history of what could have been had classic beat 'em ups not receded into the background after the shift to 3D. During an age of radical reexamination and reinvention of beat 'em ups, SpikeOut proved that the old traditions could just as well be made to work in 3D. SpikeOut’s only problem was that nobody outside Japanese arcades was around to see it, nor its spin-offs SlashOut and Spikers Battle.⁷ The only available Sega Model 3 emulator not being very user-friendly didn’t help matters either. SpikeOut: Battle Street failed to catch on or inspire a mini-revolution in the same way that DMC1 did, and neither did other 3D beat 'em ups like Gekido, Fighting Force, The Bouncer, Urban Reign, Oni, Final Fight: Streetwise and God Hand. Meanwhile SpikeOut director Toshihiro Nagoshi's new beat ‘em up franchise would stray down the dark path of RPG-ification until they started making full-on turn-based RPGs. Ironically it’s because of the newest turn-based Yakuza entry that some public interest in SpikeOut was reignited.

This is purely wishful thinking, but had Sega given it a home system port for the Dreamcast alongside Sega Rally 2 and Virtua Fighter 3(tb), and had Sega capitalized on their ownership of Streets of Rage–the flagship franchise of beat 'em ups in the eyes of Westerners at the time and even today--to market it as a spiritual successor to Streets of Rage or even reskinned SpikeOut to be the long-awaited SoR sequel released 4-5 years after the last SoR game⁸, it might have ushered a minor renaissance of 3D brawlers in the same way Streets of Rage 4 did for belt scrollers, or at least established some kind of cultural canon through which future 3D brawlers could be better understood by the public, in the same way that Super Mario 64 did for 3D platformers. Only releasing a home port/sequel on the (relatively less popular) Xbox of all consoles in 2005--a time period when understanding and tolerance by the public and professional game reviewers for arcade sensibilities was fast approaching rock bottom, as we’d see again a year later with God Hand--it was doomed to die forgotten and appear antiquated, especially when compared with the paradigm shift and complexity of Ninja Gaiden Black that was released on the Xbox that same year, or the spectacle offered by God of War or the freedom of Devil May Cry 3 on the PS2. One must imagine what the current landscape of 3D action games would have looked like had there been a proper Streets of Rage 3D.

The past couple of years the trend in 3D action games has been to lean more heavily on timing and rhythm game elements as the meat and potatoes of combat. How you move and where you stand doesn't matter as much as pressing the right defense button at the right time does. Ever since the success of the Arkham Batman games and especially after the success of Sekiro, the idea of offense and defense as rhythm minigames proved to be popular with mainstream audiences. The idea of “hit the right button at the right time” is simple to understand and feels good to execute, whereas concepts like spacing and positioning and manipulating enemy behavior are obtuse and don’t come with simple answers on how to approach them. The simplicity of timing-focused combat might be its appeal, but is also its weakness. When designing challenges for your game, you can’t do as much with only time than if the gameplay was designed to test your mastery of both time and space. This becomes a problem if a lot of developers in the same space look at what’s popular and well-liked, and decide to reuse this inherently limited template for their own new action games. Their gameplay all starts to blend together, and market oversaturation starts to kick in.

It’s at times like these that it’s worth looking back at the lesser known games, at the forgotten experiments and weird genre hybrids, to see for yourself what could have been and decide what could be. With some luck you stumble upon something like SpikeOut, and realize that there can be more to games about punching people than playing Guitar Hero.

appendix i. - asides

• Since I couldn’t find any comprehensive information about the changes between the original DBO version and Final Edition, I decided to compile a changelog myself. You can find a work-in-progress list of it here. Once I’m confident that I caught most changes, I’ll put those changes side-by-side in a video.

• Final Edition adds some extra content, like two new stages, character emotes, bonus rounds, and some secrets. But it also changes the combat in some ways. For example, super grabs are actually useful now and deal extra damage to surrounding enemies. Certain stages flow better now that they make use of progressive spawn waves instead of only spawning the next wave after you K.O’d the entire current wave. Most bosses also had their HP reduced. Plus, what most bosses would do on wake-up was usually consistent (backing away while invincible, you'll see me consistently meaty bosses this way in my own clear), but in Final Edition this was made random–you now have to rely on option selects instead of guaranteed meaties. Another key change was that the invincibility was removed on your grab side switch. It used to be that you could leverage the invincibility of side switching and jump throw startups to i-frame through entire enemy spin attacks into guaranteed jump throws (jump throws are the only non-special throw capable of knocking down bosses, so boss fights would see a lot of jump throw spam). But now? Side switching to the back into jump throws ain’t free. It depends on the enemies around you if you can safely pull it off. And to spice things up even more, some enemies can now recover mid-air from jump throws and C2 attacks, although that’s just getting kind of silly.

• Thankfully boss enemies in this game largely behave like regular enemies, and are always accompanied by supporting minions to prevent you from being able to stunlock them. There are no annoying boss gimmicks or static sidestep-then-strike patterns the likes which we’d see later in SlashOut. However, all boss enemies do have their own special set of rules to make them spicier. Many of them are resistant to being knocked down from most moves (except your charge moves, specials, and jump throws), they have access to invincible or super armor moves, and they can randomly break out of jab strings. This sounds really annoying to deal with, and in some cases it definitely is, but it’s more manageable than you think. The random hitstun breakout sounds horrible until you find it doesn’t apply to being grabbed or being stunned by a C3 (except for a few bosses who are immune to C3 stun), after which you can go jugglin’ with the boss. Mostly it just serves to make you do more commitment-heavy combos on bosses to make the most out of each opportunity instead of basic jab infinites. Hitstun breakouts being random is rather inelegant, but it works. Bosses doing invincible moves in the middle of crowds also both keeps you on your toes and reminds you to prioritize the minions for throws and i-frames. It’s only when those invincible attacks are very long (over 3 seconds) that it really gets annoying, as with the anchor and scythe bosses.

• The only strange thing with bosses in general is that the minion spawns are always finite. This means you can clear out all the mooks first and then go loop the boss in a corner, which goes against the entire point of the genre–fighting enemy crowds. I do understand why they did this, since time refills are tied to bosses spawning in, dealing enough damage to a boss, and K.O. ‘ing enemies, so having infinite minion spawns means you could milk those for score or time forever. Although I don’t see why the game couldn’t have made an exception to time refills from minion K.O. 's after getting a time refill once or twice. It’s not like it doesn’t already apply exceptions for time refills, where duo or trio boss enemies only net you time refills on K.O. instead of also taking them down to 50% HP like all solo bosses.

• I'm very certain that Sega very definitely did not mean anything by having the only enemies in the Sega-themed mall entrance with the big Sonic the Hedgehog statue in the middle be all yakuza.

appendix ii. - footnotes

---

1. In line with tastes at the time, SpikeOut: Battle Street for the Xbox would take all of SpikeOut’s vibrant Dreamcast locations and make everything look like it's set in the UK. Yuck! At the same time I'd say Battle Street's SuperSweep-powered soundtrack headed by legends Shinji Hosoe and Ayako Saso is several steps up from the original soundtrack. You win some, you lose some. I don't know what the gameplay differences between BS and DBO/FE are other than that BS has many more playable characters, including even boss enemies? Narratively it's a sequel where you play a grown up Spike Jr., but gameplay-wise it's basically a port/remake, down to reusing the exact same level geometry and enemy names and even spawn positions(?) of Final Edition, as if nothing really changed in the passing 12 years, if that makes sense.

2. In most belt scrollers every enemy type would have its own unique hardcoded wakeup time. Some would take a long time to get up after getting knocked down, some would get up immediately. This seems like a pointless thing to add, until you consider how looping a pile of overlapping enemies would otherwise look like. You’d gather them up in one big pile, knock them all down at once, they’d all get back up at the same time, and you’d repeat this until they were dead. But if each individual enemy type was to get up at a slightly different time, they are more likely to wake up during the gaps of your strings. Unable to keep every enemy in hitstun at the same time, your enemy pile would quickly fall apart. And besides, variable wake-up timing would make it harder to even round up enemies into a pile in the first place. All in all, this is pretty essential to have in a belt scroller if you don’t want a lot of encounters to eventually revolve around looping crowds to death again.

3. SpikeOut could have kept variable wakeup enemy timing, but here it would have arguably been annoying and redundant. Annoying, because SpikeOut has about 30 enemy types (not counting boss enemies) with roughly identical behavior and traits, each of which only shows up for 2-4 specific areas, all of which would pose an extra knowledge burden for you to memorize. Had it been like 12 recurring enemies it would have been tolerable, but 30 different enemies that only appear a few times is a big ask. Redundant, because SpikeOut’s more complex juggling system inherently allows you to make enemies hit the floor at different times no matter their wakeup time. Then consider how knocking over a crowd with a C4 inherently causes enemies to get knocked down at slightly different times. Usually those in the front will be the first to wake up, but adding enemy-specific wake-up times to the mix would make this fuzzy in a way that makes it too hard to read which enemies will be the ones to wake up first. So having variable wake-up timing here would counterintuitively make things easier for the player–but also more inconsistent. Arguably having to work around enemies waking up at the same time and actively preventing it from happening leads to more interesting decisions to make, than had the order at which enemies wake up been both obfuscated and generally in your favor.

4. The intent and consequence of this idea is similar to attacks having long wind-ups and thus requiring good foresight in terms of time and space to make them hit (without getting hit yourself), like what you'd see in Monster Hunter. Although with enemies in SpikeOut being as numerous and sticky and quick-to-counterattack as they are, using attacks with long uncancelable wind-ups in this context would quickly leave you dead, and even more so if you were locked in place while doing them. Being able to then move and do other things while "winding up" a charge attack that you can choose not to perform then gives you enough breathing room to use them without getting constantly interrupted. That way you can charge while doing other attacks, letting you do things like BBB-C3 or homing jump slam-C4.

5. There's also the possibility of integrating platforming into the combat via more complex environments, something that the belt scroller perspective in 2D wasn't particularly suited for due to the lack of depth perception for the Z-axis. The modern Ninja Gaidens and the PS2 Shinobi games placed a greater emphasis on using walls for movement. On the other hand the collision detection for climbing/descending platforms in SpikeOut feels like shit (approaching a ledge feels like you get teleported up/down platforms) and thankfully the game rarely expects you to do any platforming.

6. The only downside is that without overt paradigm shifts, as a spectator it might seem like you're doing the same thing over and over. Especially when an average SpikeOut clear takes well over an hour. Of course being in the driver's seat or knowing what it’s like to be in it gives you a wholly different perspective, but you have to attract people to get into the driver’s seat in the first place. In this case SpikeOut’s sales pitch of “3D Streets of Rage with online co-op” is more than enough to turn heads, but it’s unlikely that its approach to enemy design would work for sequels or spiritual successors without making them look too derivative. On top of that, SpikeOut’s high difficulty, long runtime, and lack of setpieces makes it a horrible horrible game for creditfeeding. If you don’t fundamentally understand beat ‘em ups, you are going to die constantly for an hour in scenarios that all look identical, with not even any setpieces to keep you entertained the same way something like the Dynamite Deka games could, such that no other explanation is left but to describe SpikeOut as a quarter muncher that isn’t even balanced around solo play. You only need to read some of the negative reviews on SpikeOut as an example.

7. According to a Japanese SpikeOut guide, SlashOut failed to really catch on in the arcades because on top of simplifying SpikeOut’s combat, its leveling system made drop-in co-op impossible, as anyone who joined in the middle of a run would be too underleveled. Most people would just stick to the Final Edition cabs instead. That SlashOut didn’t get a Dreamcast port is even more baffling. It runs on the same Sega NAOMI hardware as the Dreamcast, so it should’ve made porting it rather straightforward, and considering its engine must’ve been similar to that of SpikeOut, then surely Final Edition must’ve also been possible to port? Doubly when you consider that Rent-A-Hero No.1, a Dreamcast open world RPG that literally uses SpikeOut’s combat engine, came out in 2000. This gets even more baffling when you consider the release of Spikers Battle in 2001. This was a versus arena fighter spin-off of SpikeOut, using the same SpikeOut characters and controls and visuals and systems and even some old stages… on the Sega NAOMI! Then again, we could probably attribute these baffling choices to the mismanagement of Sega at the time and the discontinuation of the Dreamcast in 2001. Oh well.

8. Ancient–the original developers of Streets of Rage–did make a demo reel of possible ideas for a 3D Streets of Rage on the Dreamcast around 1999, but for whatever inconceivable reason the upper management at Sega of America or Japan refused to go ahead with the project. To this day, this decade-old gas leak in Sega HQ has never been patched up.