FYI, I passed TransparencyAttributes.SCREEN_DOOR to
GraphicsModeControl.setTransparencyMode(), which doesn't seem to have the
desired effect. It seems as if this may not be implemented in Java3D (at
least Sun's x86
implementation), because the result is a totally opaque surface. Sending
TransparencyAttributes.BLENDED, NICEST, or FASTEST yields the same, 'banded'
images that I posted yesterday.
I realize that transparent images can be confusing. However, it is
particularly nice in this case, because it allows the user to get a sense of
the relative orientation of the isosurface to the underlying molecule which
Bill mentioned sorting the transparent polygons as a possible solution. If
the (isosurface) polygon data structure is accessible from VisAD, would
implementing a simple z-sort of this list be possible? I am not so
concerned of a 'real-time' z-sort as the user rotates the molecule; a simple
z-sort at the end of rotation would be sufficient to ensure that they have a
nice looking picture after they get things into the desired orientation.
Would this be a reasonable approach, or a futile waste of time?
If this works out, I would be happy to contribute the resulting code.