> I thought that there was some general getValue(x,y,z) function in Field,
> that allowed the renderer to render volume transparent of the actuall grid.
> This makes it of course a bit more complex.
There is such a method, evaluate(), but almost all Field
rendering techniques depend on the topology of the Field
sampling. For example, iso-surfaces (and iso-curves for
2-D Fields) depend on topology.
> Why are there 6 texture plane sets, and not only 3. It was my impression
> that the texture planes would look the same from both sides. Is it to
> bypass the depth sorting problem in Java 3D or something?
There need to be two for each axis (X, Y and Z) so the planes
can be correctly sorted whether you are looking toward the
positive or negative direction on the axis.
> Resampling works, but takes away much of the resolution since most of the
> elements are concetrated around one area. What about boundry extraction? Is
> the easiest way to simply locate all triangles on the boundry and put them
> in a Irregular2DSet?
You might try providing users with interactive control of
the bounds of the Linear3DSet, so they can zoom the resampling
in on a small area.
VisAD does not provide a method for computing boundaries, so
you'll have to write one. You'd put the boundary triangles in
an Irregular3DSet with manifold dimension = 2 rather than an
Irregular2DSet. Finding them could be tricky. Perhaps Curtis
has someting to add.
Cheers,
Bill
----------------------------------------------------------
Bill Hibbard, SSEC, 1225 W. Dayton St., Madison, WI 53706
hibbard@xxxxxxxxxxxxxxxxx 608-263-4427 fax: 608-263-6738
http://www.ssec.wisc.edu/~billh/vis.html
