[ODE] Trimesh problems

Jon Watte (ODE) hplus-ode at mindcontrol.org
Tue Jan 10 18:11:23 MST 2006


You might want to look into supporting paging in your terrain engine. If 
nothing else, then because the area you want to cover is physically very 
large, and the items you'll want to simulate are correspondingly small. 
If you want to use float, you have to cut it into cells and only load 
the N most important cells at one time, and offset queries based on what 
cell is the center of attention.

Cheers,

			/ h+


Anders Olofsson wrote:
> The orginal dem/heightmap had 4470*2559 points, it would translate to 
> almost 23 millions triangles. Maybe it could be done with a heightmap now, 
> I dont know.
> 
> But there are other reasons too, I wanted to have a detailed coastline, the 
> heightgrid resolution is 30 arc seconds, about 500-900 meters, not so 
> detailed. And I also project the heightmap data (longitude/latitude -> 
> orthographic), so the heightmap is not spaced regulary.
> 
> The heightmap is pretty flat at places so after I have optimized it, 
> projected it and clipped out only the landmass (getting a coastline) its 
> only about 0.6 millions of triangles.
> 
> Basically I'm converting this dem:
> http://www.biologigrand.ac/lgh255/dem.jpg
> 
> to this mesh:
> http://www.biologigrand.ac/lgh255/shot.jpg
> 
> Maybe you can spot where on earth it is?!, I actually live somewhere on the 
> green thing. :)
> 
> But since it was orginally a heightmap some properties of the map applies 
> to the mesh. No overhangs (folded geometry) for example.. So I can find out 
> wich triangle is right below a point and then check for collision with that 
> point and the plane that the triangle lies on. Maybe that would work.. 
> Think I'll try.
> 
> --Anders Olofsson
> 
> 
> At 00:25 2006-01-11, you wrote:
> 
>>If you really do have a heightmap, you're throwing away a good solution to 
>>try to convert it to a trimesh.  Why do you say the heightmap would have 
>>been too costly?
>>Heightmaps have several advantages.  They get the collision right even 
>>with full penetration, the normal calculations are always good, and they 
>>are more efficient.
>>
>>Geoff
> 
> 
> _______________________________________________
> ODE mailing list
> ODE at q12.org
> http://q12.org/mailman/listinfo/ode
> 
> 


More information about the ODE mailing list