I'm trying to demo a dentable wall, as a concept point, and having some trouble handling the mesh. This is mainly because, rather than using a "rasterized" model that arrays the verticies, two large triangles are dented, deformed, and tesselated when impacted. The entire stumbling block is, "What kind of a tree handles triangles and sub-triangles, a trinary tree?"
The base idea is sketchy. Individual triangles would be at the end of the branches, and internal coordinates would be based on proximity to verticies (from which you CAN find rectangular coordinates) and depth. I'm guessing that this is way too complex for reasonable simulation, but does anyone have some experience or opinion on the matter?

-Sta7ic Matt

I would divide the wall up into quads and then subdivide those as necessary, in an octree-like method.

... and it's called "ternary"

Hmm. Thanks. I'll try a quadtree implimentation first, and redo the demo with a ... ternary ... if the inclination comes back.

Now, to re-re-reread the section on loose octtrees. Which brings to mind the point that there aren't any good articles on data structures on gamedev.net -- kinda funny.

-Sta7ic

No no no... I meant "ternary" is like "binary", so it's a "ternary tree"