[ODE] Using rays to generate a light/shadow map

[Shamyl Zakariya]

Hello!

I've got a terrain system  managed by ODE as a trimesh with great 
success. Further, I use ray intersections with the trimesh for 
height-for-point calculations ( to keep the camera above ground ) and 
for laser beam intersections. All this works *beautifully*.

I mention this because I consider myself to be reasonably proficient 
now with rays and trimeshes. I've read the documentation six ways to 
sunday and I thought I'd learned the quirks of ray intersections by 
this point. Of course, I readily admit I may have missed something.

Anyway, I recently thought, why don't I add static shadows to my 
terrain? I've already got a lightmap which uses surface normals and 
height factors to determine final color, so having static shadows would 
just be a matter of casting a ray from the center of each triangle to 
the lightsource and determining if any part of the terrain's geom 
intersects that ray -- I could then use this to lighten or darken my 
lightmap accordingly. Simple enough.

Well, the trouble is that it produces garbage! I know that my lightmap 
is still being stretched correctly across the terrain because lighting 
and color are still correct outside of the garbage shadowing.

So, printing out contact info for each collision I get, apparently, 
contact points just a smidge above ray origin. E.g., if my ray is 
emitted from 0,0,0 and the surface normal of that triangle is 0,0,1 and 
the normalized vector to the lightsource is 0,0,1, I'll get a contact 
point like 0,0,0.46.

Here are some real numbers, randomly selected from a dump of about 
30,000 erroneous contacts:

surface( 791.01, 4990.24, 0.57 )  <-- ray origin
normal( -0.00, 0.00, 1.00 )       <-- surface normal
toLight( 0.00, -0.04, 1.00 )      <-- vector to lightsource
contact (791.01, 4990.24, 0.59 )  <-- point of ray contact
contactDist: 0.014612             <-- distance to point of contact

For reference, I'm treating +Z as up.

Anyway, I know there's nothing for the ray to collide with. Also, I'm 
projecting the ray's origin away from the triangle center along the 
triangle's surface normal by 0.5 units so the ray won't collide with 
the source triangle.

So, the only thing I can guess is going wrong here is this: I'm doing 
this lighting calculation during the loading of a level, after the 
terrain geom is created, but before calling dWorldStep -- dWorldStep 
isn't called until the loading process is complete and the game playing 
begins.

So, I'm thinking that even though I'm manually calling dCollide with my 
ray and with the terrain geom, perhaps there's some magic that needs 
doing by calling dWorldStep at least once?

Am I missing something obvious?






Shamyl Zakariya
   "this is, after all, one of those movies where people spend a great
   deal of time looking at things and pointing."
	From a review of _Fantastic Voyage_