The expensive resource (analagous to rendering time in the LevelOfDetail
and culling cases) is network bandwidth. To reduce network traffic, the
server needs to have more information about what I am interested in, so
it can "cull" out the pieces I'm not. This might take the form of a
"detail indicator" that gets passed with my request ("show me the scene
graph down to level 5"), or it might be a pattern that I had pre-registered
with the server.
I think your dynamic web of VRML entities could be implemented easily if
browsers had the same functionality as servers. Maybe getting rid of
the distinction (just as an intellectual exercise) would make the issues
stand out more clearly.
-pete
-- Pete McCann [email protected] Department of Computer Science http://swarm.wustl.edu/~mccap/ Washington University in St. Louis