Animation strategies in 3d rts's

OpenGL Development JOGL Development
#1

I have a fairly ambitious project of mine that’s beginning to come to fruition. My graphics engine is basically complete and I’m getting ready to make actual game related decisions.

So at last my question is:
How is animation usually accomplished in 3d rts’s? Is it through skinning, tweening, or some way I don’t know?

I’ve tried googling but the results were disappointing.
For my game I’m picturing that there could be a total of 500-1000 troops in the entire battle, with 200 on screen tops (this is just a guess), so I have doubts about skinning being efficient enough even if I use shaders just because of all of the quat and matrix math for interpolation. Maybe I’m mistaken, but I’m at a loss. The game will have the potential to get reasonably close to units so the animation needs to be reasonably detailed.

Thanks

#2

in cases where large groups of the same model are animated, shared vertex skinning is used. that means, that every tenth model oder so uses the same animation than another one. so you have only 10 instances of the animation in memory although there are 1000 guys moving around.

and nowadays skeletal animation is used more widely.

#3

That seems like it could get into trouble if guys are doing enough things differently. If the group is all walking/moving it would work well. But what if they start splitting up and doing actions that aren’t concurrent with anyone else, there it seems like the sharing strategy would fall apart.

Also, I thought skinning and skeleton animation were the same, is there a difference?

#4

yes, the more different animations you have, the more animations-instances you will need.

yes, i think they are the same. i think there are two approaches to animated models:

  1. skinning or skeletal: only bones-rotations are saved in the keyframes
  2. mesh: vertexpositions are saved in the keyframes