B3D scene graph and efficiency?
Blitz3D Forums/Blitz3D Programming/B3D scene graph and efficiency?
| ||
| Maybe only Mark knows the answer to this: I assume B3D uses some type of scene graph structure to manage objects in a 3D scene. If so, do the parent nodes contain a bounding box that includes all children? Is that bounding box used in culling? I usually just tend to create objects willy nilly, only parenting them when I need to for rotation and such. Obviously, if you've got a car made out of 50 objects and want to move it then you'd want to stick all of the pieces together. But, for level geometry, would it be worthwhile to do some sort of spatial-based tree structure so that bits of the level get trivially rejected by B3D's culling mechanism? I keep hearing about reducing surfaces but a single surface with little bits all across a level results in B3D trying to render all of those bits each frame. Wouldn't it make sense to break up the surface spatially so that some of those polys don't ever get into the pipe - or are surfaces really the most important thing? I'm working on a tile based map thing and I don't want to dump everything into the render each frame if I don't have to. Right now - with one surface per texture - nothing's getting culled regardless of camera position, etc. As long as one poly on the surface is in view all of it is 'in view'. Just wondering what's most efficient. M |
   |