Hi Paul,
So I think the osgocclusionquery test case is contrived, but if you
really need it to work, then one workaround would be to change the
example to disable autocomputation of near/far. But to "fix"
OcclusionQueryNode so that it handles this case "better", we'd need to
do a full query in this situation, and I guess it's debatable as to
whether this would actually be an improvement or not.
Thanks for the explanation, it makes total sense. When the box is viewed
with the view vector perpendicular to one of the faces, the computed
near plane grazes the surface, so it intersects the bounding sphere.
I guess I would modify the example to disable automatic near/far
computation, with a clear comment as to why we do this and why it
shouldn't be a problem in regular scenes (so people don't go disabling
it if they don't need to in their apps). That way the example would
perform well and people just looking at the example wouldn't get the
wrong impression that there's a bug or something like that.
I'll try to get to this soon.
I admit I didn't design this code to work on older hardware, nor have I
tested it on hardware without this capability. Please feel free to make
any necessary changes to that it at least behaves like a regular Group
node when the feature isn't present.
OK, well the first thing I'll do is obviously try it :-) My analysis of
the code might be wrong, it's just after a quick read-through. But yeah,
if I see it's behaving wrong, I'll try to fix it.
I've integrated it successfully into our software, btw. Was very easy,
and works great (at least on this machine, nVidia 9800 GTX+). Now I just
have to educate our artist on where to place those nodes. ;-)
Thanks,
J-S
--
______________________________________________________
Jean-Sebastien Guay jean-sebastien.g...@cm-labs.com
http://www.cm-labs.com/
http://whitestar02.webhop.org/
_______________________________________________
osg-users mailing list
osg-users@lists.openscenegraph.org
http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
