On Wednesday, July 2, 2003, at 03:28 PM, [EMAIL PROTECTED] wrote:


Darrell Walisser wrote:

I don't really know how to go about fixing these things, I'm just
reporting on what I think needs work - not that this needs to be done
now, of course (optimization comes last in my book). I just anticipate
it becoming more of a headache in the future when we try to increase
scene complexity. Of course, it will need to be fixed to be able to
compare visually with GPU-optimized sims.

One thing I was thinking about was minimizing the number of vertices sent to the GPU by using what we already have.

Well, you could do that, but I don't think it will help much. The scene itself doesn't really have that many vertices to begin with (I think normally less than 7000, IIRC). LOD will help increase detail with minimum performance impact though (but we
ll still be seriously CPU-bound). The source of the problem isn't the vertex count, it's the way vertices are processed and sent to OpenGL (which is controlled largely by how tiles are created).


However, I think a good first step would be optimizing what we have already, then think about LOD to get even more detail nearby. A place to start is the part in TerraGear where the trifans are formed. Here you would probably form tristrips instead (since you can cover larger areas), and have a constraint that tries to get a minimum number of vertices per chunk (for optimal rendering), while at the same time another constraint that tried to keep the vertices close together (for optimal clipping). You'd probably have to do this after the land usage/roads/etc were created to get the best result.

I was interested in this problem a long time ago and I found some papers about stripification of irregular meshes (drop me an email)... from what I remember, it is very difficult to do. It depends on how flexible the scenery tools are, which I have no idea about yet (will changing fans to strips bork something down the pipeline, in otherwords).


_______________________________________________ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel

Reply via email to