What could happen is you have the description of the object, which
includes things such as what kind of object, what material, and
interactions. The client then gets its best local representation of
this, or loads the original, based on defaults as mentioned
before. For higher level rendering, the material is as important as
the object itself, defining reflectivity, etc. Just as an object would
have material(s), a material would have texture(s), which could also
be best-fit. The usere may not wish to download "reallyhuge.gif"
whenever they go to a new place, so just defaults to something.
(although such files are often rather crucial in material building)
It becomes a question of at what stage does substitution of one
object/material/texture for the original change the meaning of the
original scene?
Ray