[ODE] Re: Heightfield / Collision.

gl gl at ntlworld.com
Tue Feb 18 23:58:02 2003


> I havent read further yet but I can see the point here. The effect would
be slight but depending on the terrain noticable, especially if the designer
was trying to create some effect by having the object on the hill hit an
object below. Sure the askew path would be repeatable.
>
> I guess a small 9*9 beizer patch with the heightfield as the control
points is out of the question :)

First off, I'm not sure this is a problem (maybe someone who's done this
type of work can comment).  But anyway, it seems to be that what you're
asking for is more accurate _normals_, not more accurate sample heights.

This all depends on which camp you fall in of course (see my other mail) -
but there may be something to interpolating normals even for the game crowd
(although with the callback approach the app can chose how it calculates
those things anyway).  Let me get spheres working in a vanilla ODE way, then
we can all experiment with it and see.
--
gl