I'll be making a posting on caching, probably tomorrow, but I wanted to get
a few things clear, before I begin, and these questions point up the real
reasons we need a caching architecture:
>No repository or CDROM could possibly contain "enough" models of
>"common objects" (real and imaginary, I presume). And where's the
>originality in scenes made up of the same 3D models?
Most all of the real world is made up of a few hundred basic shapes,
endlessly repeated. Light bulbs. Stairways. Ceiling tiles. Electric
outlet wall-plates. And so forth. If basic shapes can be scaled and
wrapped with textures - both of which are easy in VRML syntax, we can create
a world at least as rich as the one we inhabit day-to-day without incurring
huge bandwidth hits.
>3D models are valuable property today. Collecting, describing,
>converting and indexing thousands of models will be a lot of
>work.
That era is probably at an end. The doors on 3D have been thrust open wide;
in much the same way that people trade desktop wallpaper patterns or dirty
GIFs today, you'll see them trading 3D objects around in 6 months. Seems
that on Internet, people really do like sharing; it's both gratification for
work well done, and self-promotion. With applications like TrueSpace and
VHSB just around the corner, everyone will be making interesting 3D objects
to trade and share.
The market for 3D objects will probably migrate to the high end.
>Who's going to be the clearinghouse?
Perhaps a non-profit organization such as W3C (which has already agreed to
serve in this capacity, but that's another posting, too), or some
equivalent. This is not without precedent. The RFCs are kept at
rs.internic.net, along with the master DNS tables - a rough analog.
>For example, take
>a look at your copy of the Viewpoint DataLabs catalog, certainly
>the largest collection of 3D models available today. Even still,
>does it contain all the models of "common" objects you'd want?
Sadly, it does not. It's probable that more 3D modeling will be done in the
next year than has happened in all of previous 3D graphics.
>Are you asking for 3D versions of everything conceivable in the
>Popular central repositories will run more slowly, will be restricted
>if over-used, or might be down, not to mention wasting time for another
>connection, and exist at the beneficence of someone willing to put
>gigabytes online at high speed.
They should be mirrored. Another reason for an abstract notation of what an
object is; URLs differ from URNs in that URNs are abstract and can point to
different servers for different people.
>Most world-builders will want absolute control and assurance of the
>models in their scene, and will avoid techniques based on slippery
>sources.
For commercial purposes (say that IUMA does "Madonna World") this is true,
but for the average AOL or Prodigy or CIS user of VRML, that's unlikely.
They'll want to make their world accessible, otherwise, like people who
install 500K GIFs on their home pages, pretty soon no one will come to visit.
>All this effort, for what? For people to save local hard disk space?
To save a huge amount of transmission time; even with binary formats (which
we need immediately) a rich environment will take time to send down the
wire. Every possible way we can optomize that, we improve the usability of
VRML and enhance our own chances for success.
>If freeware object collections arise, they'll be on the net or CD.
>You can make a copy of what you want. You can color it however you
>like, scale it, decimate it, whatever. You can make it *your* teapot.
These are not mutually exclusive conditions.
Mark Pesce
VRML List Moderator