Zorba the Hutt (zorbathut) wrote,
Zorba the Hutt
zorbathut

Dead Space Dissection: The Trouble With Ragdolls

Traditional game animation is (mostly) pregenerated. An animator sits at a computer and carefully poses the motion of each limb. Eventually, you have a spider that crawls across the ceiling and shoots acid in your face. Done!

In motion, this looks pretty good. In death, it’s problematic. First, unless you’ve gone to the trouble of multiple death animations, creatures always die the same way. If you kill five hundred Basic Guards, you’ll end up with five hundred identically-posed corpses lying around. Uncool. Second, death animations have nothing to do with the weapon you kill them with. Poke them a thousand times with a needle? He’ll scream, fall over, and lie with his face on the ground. Shoot him with a portable nuclear warhead launcher? He’ll scream, fall over, and lie with his face on the ground. Uncool. Third, death animations tend to “snap” from other animations. Basic guard takes a flying leap, jumps at you, you kill him midair . . . and suddenly he plays the Death Animation, which involves him instantly standing up in midair, then screaming, falling over, and lying with his face on an imaginary ground, while his corpse eventually falls into a pit. Uncool.

There’s a solution to this. Games have gotten sophisticated enough that most modern 3d games include a basic physics engine. You don’t need perfect physics for this, something simple is pretty effective. Animations are already based on a simple skeletal model – arms have two “bones”, legs also have two “bones”, etc – and it’s easy enough to allow these bones to just move via the laws of inertia and behave properly on impact.

So you kill someone on a tower of boxes, his corpse will tumble down the boxes. You shoot someone with an air cannon while he’s standing in front of a railing, he’ll backflip over the railing. Rocket launcher to the feet? Flying guard corpse! Cool.

There’s problems. (Of course there’s problems. You think I’d be writing about it if it really were that simple?)

Ragdolls tend to be used only for actual death. It’s just too hard to recover from a ragdoll collapse if the creature isn’t actually dead. You knock a Basic Guard into a pile of boxes and he gets jackknifed between two – how does your Basic Guard recover from this? He doesn’t, but now there’s a living Basic Guard jammed uncomfortably into a pile of boxes. It doesn’t work well. So ragdolls are only used for death.

But that introduces a new, irritating problem. Ragdolls can be used to detect death. Dead Space includes a gun that fires a shockwave which knocks things down. When knocked down, a lot of the zombies will cheerfully play dead, only to eviscerate you when you turn your back on them. However, it’s trivial to determine if they’re dead or not. See, when you knock them down, they always fall on their backs, with their legs facing you, and their left leg (from your perspective) slightly lower than their right leg. I know this very well from knocking down dozens and dozens of zombies this way.

When I see them fall down this way, I know they’re just going to get up again in a few seconds. When I see them fall down any other way, I know the ragdoll mechanic kicked in, therefore I know they’re dead, and therefore I can forget about them.

It’s not very suspenseful.

I’m not sure what the solution is. It really is incredibly hard to recover smoothly from a ragdoll-based collapse. On the other hand, unless you have your artists make dozens of death animations, it’ll always be easy to distinguish a “real” ragdoll death from a “fake” non-ragdoll death.

But it’s a problem, and in a game like Dead Space, where detecting Proper Death is a very valuable skill, it’s distracting like you wouldn’t believe.

Tags: art, design, dissections
Subscribe
Comments for this post were disabled by the author