There is text that behaves in that manner in at least one VRML model that
came with the WebFX plug in for Netscape. It looks gesat. I'm not near
my PC and can not remember the name of the weather satelite VRML model.
I would love to see the capability you described "a single polygon object
to rotate on an axis >(usually Y) so that it always faced the camera."
This would especially be useful if each rotation of the polygon would
(optionally) permit the attachment of different texture map image.
To make this capability even more useful would be the ability to rotate
the polygon around a point in space. For each (angle within a given
range)(optionally) a different polygon would be appear with a(n
optionally) different texture map or color associated with it. This would
permit the rapid display of an irregularly shaped and colored model object.
Add LOD to this and you can switch to a traditional VRML object (i.e.,
a building) when close and perhaps go inside of it.
While I'm rambling... add a time component to the single polygon idea
in a VRML model and there exists the ability to see a model change. A
building or city could rise out of or disintegrate into a flat expanse
or another complex model.
With the benefit in reduced viewable images to render will come the
burden of determining which polygon and associated image/color to render
as well as reloading candidate images to be viewed.
I have been building several complex VRML models from CAD generated
geometry and am disapointed at the lousy performance with several
PC-based VRML browsers. There is too much geometry. If I could use
the (series of) single polygon(s) method and attach a rendered image(s)
to it I believe performance would be dramatically incesased.
John Chapman