3D Animation

I Wanna Rock!

January 6, 2009 January 6, 2009 By now, you’ve probably heard about the phenomena that are the music video games Rock Band and Guitar Hero. I can’t think of any experience that has done more to broaden the appeal of video gaming for new players. Given their wild popularity – and my own fondness – I thought I’d take a look at the latest versions from the perspective of a 3D Animation student. Both Rock Band 2 and Guitar Hero: World Tour are available on current- and next-generation consoles, but definitely look best when played on one of the newer systems. My experience has been on the Playstation 2 and Xbox 360, but I would be surprised if there were significant differences on the Playstation 3. Animation in the two games is very good. The developers have created a library of what I like to think of as “rock cycles” that they string together to create a compelling performance for the onscreen characters. There are probably 8 or 10 cameras used during any act, and cuts between them are well timed. My guess is that the animators develop a performance for each song using their library of cycles and a basic skeleton with rig. This allows the player to create and use whichever avatar they want; then the avatar is applied to the skeleton when they play. The lip sync is also well done, although I think Rock Band 2 did a little extra work on their phonemes and visemes. Guitar Hero: World Tour has a lot of rapid head movement and often hides the singer’s mouth behind the microphone. Textures in each game are gorgeous, although I enjoyed what GH:WT had to offer more because of some stylistic choices the developers of RB2 made. In RB2, a noise filter has been applied to all the cameras, giving everything a grainy look. On a standard definition TV, this gives the effect of having been recorded on film, which is nice but in high def, it just looks like static. Image by Hilary Goldstein. While the GH:WT textures are cleaner, the characters suffer from some strange issues that I think are related to texture baking. This process renders a character’s lighting information in advance so it doesn’t have to be dynamically determined by the game console. It works great if the model’s light source won’t be changing dramatically. GH:WT didn’t apply it universally – just to certain highlights.  Unfortunately, this means features like teeth, eyes and other body parts appear to glow when the lights swing away from them. RB2 appears to have avoided baking, so the color schemes in that game are consistent. Placement of lighting is better in RB2, though more organic in GH:WT. RB2 characters seem to have invisible lights that follow them around via parent constraints and provide some really nice underlighting and highlighting. Both games use some kind of volumetric effect on their spotlights, which greatly improves the mood of the performances. Image by Chris Roper The crowds are very simple repeating models, maybe even some kind of advanced sprites with 3–5 animation cycles, so it’s fairly easy to pick out identical fans that are doing the same thing. RB2 either has more animations than its rival or places the fans more carefully; the crowd seems more fluid and natural. GH:WT fans will often appear right next to each other, like twins or triplets dancing to the exact same beat. This makes me think they’re being dynamically generated by a randomizer that hasn’t been programmed to space out any duplicates. The next time you play a video game, think about how what you’re seeing was developed.  It might give you some useful ideas on how to handle your next animation. Or (even better) it might show you what to watch out for.