The Death of Rigging?
March 2, 2009 Page 1 of 3
[Originally printed in Game Developer magazine's August 2008 issue, this article from Valve veteran and Bungie staffer Steve Theodore examines an alternative to the traditional animation rigging system currently used throughout the game industry.]
No matter how long you've been out of school, August always makes you think about reinventing yourself. All those back-to-school wish lists -- new clothes, new backpacks, a chance to improve your grades or upgrade your social status -- leave a mark that never quite fades. It's a good time of year to break out of a rut.
If you're looking for a good rut to start with, think about animation. Animation is the absentminded professor of the game art disciplines: highly technical, but rather stuck in its ways.
Certainly, nothing has come along to revolutionize animation in the way subdivision sculpting has changed modeling, or Shader Model 3.0 has changed texturing and effects.
Animators are still keyframing away with only incremental improvements to the tools that debuted with 3ds Max and Maya, back when the Dreamcast was the hot gaming machine and gas cost $1.27 a gallon.
The Animator's Frenemy
The guardian of the status quo in animation is the character rig. The animation rig is really the lynchpin of a studio's entire animation effort. It's the animators' main UI element. It provides the engine with specialized markup. And it is the backbone of asset control. Most importantly, the things your rig does well (or does poorly) subtly flavor every aspect of your work.
Unfortunately rigs are -- let's be frank -- a pain in the butt. Even the best rigs are complex, touchy, and hard to maintain. Keeping track of the rig through its evolutions and keeping different files in synch is a nightmare.
No single rig can satisfy all animation needs equally well. Some rigs bury themselves in layer upon layer of features until they feel (and perform) like Rube Goldberg machines, while others, hoping to stay lean, become virtual straightjackets. If we're talking about reinventing things, this might be a good place to start.
Only masochists want to go back to animating directly on FK bones all day. But what if you could keep the good parts of a complex modern rig -- helpful UI, the right control spaces, and efficiency -- without the management overhead or 60 million control nodes cluttering up your scenes? Would you be interested?
Well, there is an alternative to complex rigs, one that tackles many of the problems we ask rigs to help with from a very different standpoint.
Like An Onion
Animation layering is a workflow that builds animations up out of layers in much the same way Photoshop assembles a bitmap image out of bits and pieces. Instead of forcing you to plan ahead for every possible contingency in the design of an omni-competent do-everything rig, layering lets you slap together whatever techniques make sense for a given shot. It can be a very compelling alternative to the standard way of doing things.
Layering evolved quietly outside of conventional animation pipelines. It's a key tool for teams that rely heavily on motion-capture or simulated data, but it's also starting to find favor among traditional animators.
The eternal struggle between the proponents of mocap and of hand animation is written deep into games industry lore, so we don't need to revisit it here. And in any case, the fact that layering evolved in response to dealing with mocap doesn't mean that's all it's good for. It is, though, a good way to understand the essence of the approach.
Whatever you think about mocap, everybody agrees that it's a slog to work with. Conventional animators' eyes roll back in their heads when they open a graph editor and see, instead of cheerful colored lines, densely packed keys marching like army ants across their screens.
The drawbacks to such dense data are all too obvious: It's slow to work with, and fixing it in place is almost as bad as returning to the bad old days of animating one frame at a time. And of course, the data is all FK.
Aesthetics aside, very few animators really embrace mocap for the sheer joy of working with thousands of keys. The need to tame that mess is what gave rise to animation layering.
The arms race between mocap and traditional keying is almost a replay of the old battle between bitmap painting and vector illustration in the 2D art world. Back in the days of the Super Nintendo Entertainment System and the Macintosh II, bitmaps were for photos or scanned clipart; large format illustration was more often done with vector drawing programs like Illustrator or CorelDraw.
Vector art made reshuffling, resizing, and replacing pieces infinitely simpler than chopping up bitmaps -- until Photoshop 3.0 introduced a generation of artists to layers.
Photoshop layers didn't miraculously end the seesaw battle between bitmap and vector illustration programs, but they gave bitmap packages a gigantic boost in flexibility and freedom.
Being able to move pieces non-destructively, to transform parts of the image while leaving the background intact, and to experiment turned Photoshop into the all-purpose juggernaut we know and love today, while vector programs were gradually relegated to technical illustration and graphic design.
When layers first came out, they were a geeky curiosity, but it didn't take long for them to become a solid pillar of the illustrator's toolkit. Will the same thing happen in animation?
Page 1 of 3