Re: [Algorithms] Leafy BSP Tree Contruction

[<chr...@pl...>]

Manolache Adrian wrote:

> I checked and there are a lot of cases when vertices are very close to
> a lot of planes, and this causes problems when splitting with these
> planes.

If the vertex is sufficiently close to a plane it should
be considered ON the plane; this is what it meant by a
fat/thick plane.

It sounds to me that you might not be correctly handling
the thick planes.

The introduction of thick planes changes not only how you
clip/split polygons to the plane but also how you classify
polygons w.r.t. being in front/on/behind the plane.

You can find a short outline of the problems in my GDC
lecture on robustness:

http://realtimecollisiondetection.net/pubs/

There's a lot more detail in my book on both splitting
and classifying to thick planes (pp 364-373).


Christer Ericson, Director of Tools and Technology
Sony Computer Entertainment, Santa Monica
http://realtimecollisiondetection.net/blog/