VRML and meshes

Scott Nelson ([email protected])
Tue, 23 May 1995 19:36:41 -0700


We'd appreciate any feedback as to whether a computational
mesh MIME type should be a world/* object or something else.

Although VRML could support a mesh, the classic case of
a 50x50x50 group of cells is more than a generic VRML viewer
can handle and requires a number of hints about the mesh
(a computational mesh contains a large number of simple objects,
many of which are hidden. A typical VRML scene contains a
few number of complex objects most of which are exposed).
Or enough intelligence to look at the 125,000 elements and
notice that they are arranged in a particular pattern.
These are finite element or finite difference type 3D meshes
and typically contain hundreds of thousands to millions of
elements, only of which tens of thousands of polygonal surfaces
are visible at any one time.

We are at the point now where we are either going to put
QvLib into our existing mesh readers or create a new
MIME subtype. Unfortunately, even with compression,
VRML is too generic (and large) for efficient transport.

Is a mesh a "world" ?
If so, we'd put together an RFC for something like:
world/mesh
world/regular-mesh
world/constant-mesh
world/constant-regular-mesh

which uses the subtypes as the hints. world/mesh is the most
generic and can read the other types. These other types utilize
the kinds of mesh viewers that exist presently. The "constant-" subtypes
imply meshes that *do not* change with time. The other mesh types can
change with time. All of the above can have data that changes with
time.

Time is the big element missing from VRML which makes it
unattractive right now for these kinds of applications. Our
data and our meshes change with time. This would be
similar to a VRML house deforming during a tornado or
where one object collides with another (in a 3D mechanical
simulation).

Any comments would be appreciated.

Scott Nelson
Lawrence Livermore National Lab

-- 

+----------------------------------------------------+ |Scott D. Nelson B131 Rm2074 3-1250 | |Lawrence Livermore National Laboratory | |7000 East Ave., L-153 Livermore CA 94550 | |email: [email protected] http://www-dsed.llnl.gov/ | +----------------------------------------------------+