I have three memoryblocks: one for the current frame, another for the
next frame, and another for the interpolatedFrame. The math is
simple: interpolatedFrame=currentFrame+(nextFrame-currentFrame)
*percentTweening
I'm not using RB's trimesh, I'm modifying the points and normals
directly via the API. Would RB's trimesh run faster?
A 3,000 triangle model runs at about 125 FPS. In RB 5.5, it ran at
about 160FPS.
Yowza, that's a lot of points. How are you going about
interpolating vertex normals? I asked a question about
interpolation on this list a while back and Joe talked me out of
interpolating normals and letting RB do it -- you might want to try
it and see if you gain any speed back.
In my own tests with RB2006, I've seen slowdowns with multiple 700+
vertex objects (I think I can get about 7-8 of them in there before
a see things start to slow down, and I'm only interpolating
vertices). My machine is kinda weird because I have a 1.8Ghz G4
(upgraded the CPU), so your engine runs about 12-15 FPS for me.
As a side note, morphing does work fast on multiple objects if your
objects aren't built up with too many vertices. In a morph demo
I've been playing with, I had 50 birds (36 vertices; very low poly)
flapping around without much of a performance hit.
_______________________________________________
Unsubscribe or switch delivery mode:
<http://www.realsoftware.com/support/listmanager/>
Search the archives of this list here:
<http://support.realsoftware.com/listarchives/lists.html>
|