Re: [Algorithms] BSP tree, or what?

[Jon W. <hp...@mi...>]

Megan Fox wrote:
> I'm all for occlusion where it makes sense (big city buildings, big
> houses, large walls), but most of the places where occluders make
> sense are probably also places where individual objects are big enough
> to just be modelled with occluders built in - no need for fusion.  The
>   

Think of Manhattan, when your viewing angle isn't exactly 90 degrees to 
the buildings. There will be potentially significant amounts of people, 
cars, street signs etc that are occluded by some combination of 
buildings, but are not occluded fully by any single building. This is 
especially true when using conservative or "convenient" bounding volumes 
for occlusion testing (like, spheres with a center at the feet :-)

I wasn't suggesting fusion in the sense of trees in a forest, as I was 
thinking about an urban scene. I actually think that most tree trunks 
are either too narrow, or too bent/complex in shape, to make for good 
occluders -- only the densest ground-hugging fir trees (a la christmas 
trees) would qualify. Your observations are very valuable to call me on 
that and make this clear!

Sincerely,

jw