> but what about the fact of the same 3d model is used for all the players (with face > modifications), plus the referee is not shown in vs2 series, and it has allways the > same background, for example scud race has more than 30 cars, and a lot of detailed > backgrounds, and is the most solid game of all, maybe cause it was the game with most > work on it?
Doesn't matter. The 22 player models aren't all in the same pose, so the fact that the source model data is identical is irrelevant. They have to be processed as 22 individual high-poly characters, and the animation processing is by far the most expensive part. And cars are much cheaper to animate than people. There's no bone hierarchies, no inverse kinematics to calculate, and no motion-capture data to parse. Any half-decent matrix stack setup makes spinning 4 tires trivial. Having seen Scud Race run in wireframe in an emulator, it also has far fewer polys per car even in the close-up view than the players in VS2.
The proof is in the pudding: Scud Race uses less than 25% of the available power of the PowerPC CPU in Step 1.0 (SuperModel currently gets its performance via what MAME calls "idle cheats", so games that run slower mean they are idle less). VS2 uses nearly all of it.
|