Art of War: Animating Realistic Sword Combat
December 13, 2012 Page 1 of 3
This article -- a compilation of two viewpoints on adding authentic sword combat to games -- originally appeared in the December issue of Game Developer magazine. You can subscribe to the print or digital edition at GDMag's subscription page, download the Game Developer iOS app to subscribe or buy individual issues from your iOS device, or purchase individual digital issues from our store.
Eben Bradstreet on Real Combat
The first time I walked into John Clements's Iron Door Studio, I learned how to hold a longsword.
I thought it was obvious: Grip the handle with both hands. That's how I'd always imagined it was done. It's how they did it in the movies, after all. The handle is the comfortable bit between the pommel and the cross-guard. It's large enough to accommodate both hands. So I hold it there, right?
Wrong. The right hand (or the leading hand) indeed goes just below the cross-guard. The left hand should grip the weapon by the pommel -- that's the knob at the end of the handle -- in most circumstances.
At first, I was a little dubious. "Hold it there? Really? I thought that part was just for smashing skulls. Or balance. Or decoration."
My mind drifted to Orlando Bloom in an early scene in Kingdom of Heaven, where it's quite clear that he grips his weapon in the way that seems most harmonious: by the handle, with both hands. Later in the film, he even helpfully confirms my bias by smacking someone in the noggin with his hand-free pommel.
Figure 1: The proper way to hold a longsword.
The reality of the pommel is a little more complicated. It is used to knock sense into your enemies, it does affect the balance of the sword, and sometimes it's even pretty to look at. But, it's also a great place to hold the weapon. And it's a perfect example of how all my assumptions about the sword were challenged when I first set out to learn the reality of the weapon.
Grounding Fantasy in Reality
As it turns out, how to hold the longsword is also a great place to start talking about what that reality can mean for animators. If a video game character grips the pommel with its trailing hand, the resulting animation will have fewer problems with deformation around the wrist, less clipping between the sword's mesh and the character's mesh, and will display better biomechanics when cutting (which we'll talk about later).
The first two points are subtle improvements, and are best demonstrated by gripping the handle by both hands, then extending the sword forward, holding the weapon at eye-level. From this position, start turning, windmilling, and cutting with the weapon while keeping it in front of you. If you don't have a sword immediately available, you can do this with any wooden or plastic dowel. Just hold the dowel roughly three to five inches from the end in order to simulate the exposed pommel.
As you swing, pay attention to how often the pommel wants to intersect with your wrist, especially when you try to drop the blade to the lower right, and note that in some positions, you can't continue an arc because your trailing wrist simply won't contort enough to facilitate the movement.
As I said, these are subtle points. The real magic happens when you now try the same thing, but grip the pommel, instead of the handle, with your left hand. The first thing you'll notice is just how much more leverage you have. The sword is a lever, after all, and your leading wrist is the fulcrum, so it makes sense that the further back from the fulcrum you're able to grip, the more control you'll exert on the blade. You should also notice that this method is easier on your wrist (which will help with deformation), especially if you allow the pommel to slide and turn freely in your palm. Your wrist can now stay relatively straight through most swings, and there's no longer any danger of the pommel clipping through the wrist's mesh on your character models.
Bad Sword, Bad Reference, Bad Animation
Inevitably, you'll want to capture some reference. Even if you're using motion capture, it's always good practice to "feel" the movement yourself, or to whip out a camera and go through some of the motions.
The best way to not get good reference, no matter how you decide to hold the sword, is to use a crappy weapon. For most people, the easiest access they have to a generic sword is either through a catalogue, a renaissance faire, or even the odd novelty store. With the rare exception, the weapons you get from these sources are universally terrible for your animations -- they're often much too heavy and poorly balanced.
We have one of these weapons at my workplace, and even after two years of swinging steel as a martial art, I still can't do anything with it. That weight transcends physical reality, and every skilled attempt to mitigate it directly influences how our characters move. It's an awkward and sluggish prop that makes our characters look equally awkward and sluggish. Early on, we decided not to use it.
Instead of trying your luck with weapons that will harm your animations, it's almost always better to simply go to a hardware store and buy a wooden dowel, or a length of weighted PVC pipe. If you're feeling crafty, you might even want to make your own wooden sword (historically called a "waster"). Wooden props like these will certainly be much lighter than steel, but they'll be easier to swing in a good way (good for the director, the animator, and the actor). The alternative is a poor knockoff that will cheapen your results by virtue of your trying to use it, rather than just hanging it up and staring at it (as it was made for). Bad swords make your job more difficult than it has to be.
The weapons that John and I are using in these images are made by Albion Swords (www.albion-swords.com). There can be long waiting times for these weapons, so it may not be an ideal solution for gathering good reference quickly or cheaply.
All About Universal Biomechanics
As I learned more about the art of fighting with the longsword, as Medieval and Renaissance Europeans understood it (and actually chronicled in dozens of study guides), I began to understand how intuitive the whole skill set actually was. There are a few basic guards, roughly nine vectors of attack, and a handful of rules that guide your footwork. The more advanced techniques, while impressive to look at, are largely ancillary: In all the sparring matches I've seen, the flashier techniques are never used. In fact, the more I watched and the more I learned, the more it started to feel familiar.
That sense of familiarity didn't come from the movies or the stage -- and certainly not from the highly sportified world of foil fencing. No, the moves I was seeing in sparring matches, which were reflected in the historical imagery plastered on the walls around me in John's studio, looked more like the close-quarters combat training we'd done while I served in the Army, or mixed martial arts matches on TV. It was savage and in-your-face. There was nothing at all pompous or chivalric about it. This stuff was real, and it was universal -- because no matter what time or place we come from, we're all human and we're all governed by the same biomechanics.
Consider the weapon we've been talking about. If I had been handed this weapon three years ago and someone told me to swing it, I would have done what pretty much anyone would do. Maybe swing it like a baseball bat, or because I used to cut wood when I was kid, I might swing it like an ax. If you were handed that weapon, what would you do? You might try to mimic what you'd seen Conan do, or imitate a samurai.
What you wouldn't do (at least, what most of us wouldn't do), is swing it like a golf club. But why not? Both are roughly the same length, both are used to hit things, and both do most of their business up to five inches from the end. Certainly, to swing a sword like a golf club and connect would be devastating to your target. So why don't we use it that way?
The answer is biomechanics and perception. Biomechanically, it's not efficient: The target of a golf club is at ground level, where the target of a sword is at eye level. Because the difference in targets is so drastic, we instinctively perceive that to swing a sword like a golf club is wrong.
Now let's use this thought process to delve a little deeper: Think about the target of your animations, and where your character's enemies are located. Is swinging the weapon like a baseball bat the solution? Consider the game you might be working on: In an environment where your character is trying to kill people, the target of your swing is more likely to be the pitcher (in front of you), than the ball (next to you at the moment you swing). Is it still correct, then, to stand like a batter, and swing as though you're trying to hit the ball? Probably not.
So now we can take a step back and ask ourselves: "What is the best practice?" Nothing beats calling in an expert, of course, but even if you don't have the time or resources for that, putting in a little effort in developing your fundamental understanding of human biomechanics with respect to weapons can help clean up your animations in a major way.
Page 1 of 3