Andy Ross writes:
> 
> Erik Hofman wrote:
> > There might be one step in between here, which I have been thinking of
> > a bit. It would be easy to implement a bounding cylinder (2d collision
> > detection) and only if there is a hit go to the bounding sphere.  For
> > me it looks like that approach would be much less costly as directly
> > determining a bounding sphere.
> 
> The truly adventurous should note that the "mostly 2D" nature of the
> problem lends itself nicely to a tree-based (quadtree or BSP)
> hierarchical representation.  The coarse intersection of bounding
> spheres should be the next level of refinement, finally followed by
> testing of individual polygons.

Currently all sub elements of a flightgear scenery tile are loaded
in to a 'wide' graph.

ie 

                       root
________________________________
|             |               |                  |             |  
                        Tile Node(s)
                              |
      ______________________________
      |   |   |   |   |   |   |   |   |   |   |   |   |   |   |  |
            Fan Nodes all at eaqual depth

And what IMO is needed for *fast* collision detection 
is a 'deeper' graph


                      ROOT
                          |
____________________________
|           |             |              |             |
                 Tile Node
                         |
     QUAD TREE of Fan Nodes with 
  assosciated SSG bounding spheres

This could be implemented as a load time optimization
step

Cheers

Norman


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

Reply via email to