Drawing Out the Fencer: A Design's Evolution

After starting fencing classes in the summer of 2004, it wasn't long before I desired a video game expression of the same.  But though there are scores of fighting games, none had that light, fast, to-and-fro flavor indicative of fencing.  So, I fired up the compiler and spent the next year designing and prototyping a game that would accurately capture that feel.  What I thought was going to be a simple and straight-forward fighting game blossomed into one of the most deliberate and practicality-driven game designs I had ever done. 

The game began humbly, with 2D stick figures controlled by arrow keys and a few attack buttons of varying height and commitment.  Instead of a pugilist's speed versus power tradeoff I chose a reach versus recovery tradeoff:  a deep lunge will quickly invade your opponent's personal space, but if you miss or are parried, you are a sitting duck.  But like the original Prince of Persia, the simple control scheme didn't allow for parries with any reliability.  The game quickly devolved into a pixel-perfect placement exercise with "machine-gun" stabbing.  

Interestingly, foil fencing has the same problem.  It solves it with a rule called right-of-way:  once an opponent has parried your attack, his riposte will have priority over yours even if you strike him first.  As a game feature it encouraged parrying, but it still didn't cure the unreliability problem.  A player would regularly want to parry but successfully doing so was more guesswork than anything.  Instead of machine-gun stabs, the game comprised baiting the lunge.  

This was when the design began in earnest.  I reexamined the subject matter for clues on how to fix it, but since the unreliability problem didn't exist there, I brought the tropes of the fighting genre into question.  I realized one thing quickly:  no other martial art shares the many issues peculiar to foil fencing.  

For one, all attacks are three foot long stabs, no more and no less.  Usually, swords increase one's attack range with no loss to existing options:  an armed man in close can still cut your wrist or draw-press the edge into your torso.  But a sword without an edge has very few close techniques, so a missed thrust usually means backing up or cocking back the arm until the tip is again between the fencers.  What results is a far-away tickling game with the occasional bum rush, the latter of which necessitates precision with an awkward down-thrust.  The game reflected this only so far as reacting to a missed lunge, the most obvious and committed maneuver it had.

The solution was something I would never have added spontaneously, because in any other 2D fighting game it comes across as gratuitous, sprite-blurring cinematics:  the ability of the camera to zoom.  With this unplanned feature, no longer was the lunge the only visible attack.  When the fencers were too close for lunges, the camera would zoom in so that even quick half-thrusts were plainly visible.  Now the fencers had the long distances they required for the more mobile maneuvers, but the details when in close would still remain visible.  The feel of the game improved remarkably, felt as if it finally had some breathing room.  But with this increased ability to breathe came freshly apparent peculiarities. 

Fencing has no provision for brute force, either offensively or defensively.  The swords, called foils, are a whip-like cord of metal so flexible they can be tied in a granny knot.  They are too insubstantial to push or deflect anything except paper airplanes and other foils, and the latter only by using the forte, the thicker, back half of the blade.  With such a limited ability to redirect attacks, fencing becomes a millimeter-precise geometric dance.  Touching your opponent's thrust on its foible -- the weaker half of the blade -- changes his attack angle by a mere degree, but that is enough to send his tip sideways by a critical inch.  Trying to force through your block only causes his foil to bend, sending his tip even further astray.   Speed is at best secondary:  I could not touch my seventy year old fencing instructor with any attack I cared to throw at him, because his every move was pathologically economical.  He didn't parry or attack so much as shift in place.  

By contrast, with unarmed martial arts there is little difference between, say, two consecutive left hooks that don't travel through exactly the same space.  The angle of impact, the part of the body struck, and where in the hook's power range the impact event occurs, are the overwhelming factors.  The result is that martial artists learn distinct, discrete attacks:  jab, hook, uppercut.  And though blending the basic attacks into, for example, a 30º angled "upper-hook" is possible, it isn't consciously done in practice for a couple of reasons.  One, due to the way the human nervous and muscular system works, the oft-practiced technique is both stronger and faster.  And two, misalignment of the offensive limb will cause as much if not more damage to the attacker than to the defender.   These discourage variation.

Design-wise, a small set of discrete attacks works well on a small set of discrete buttons, and this is the direction fighting games take.  But when precision is required, analog controls are the obvious answer.  Analog would have been vehemently opposed by myself at the start of this project.  If a zooming camera in a 2D fighter is gratuitous, analog controls are utter blasphemy.  But after the camera showed me the sword work in such detail, I had to acknowledge what I was seeing.  Believable fencing was a problem of input precision.  Unfortunately, two problems came with that, and one of them looked to be a design-killer. 

[Read the second half here]