Re: polygon question

Chris Laurel ([email protected])
Thu, 9 Nov 1995 13:22:21 -0800

Cindy Reed writes:
> Since I opened this can of worms, here's the perspective from a lowly
> user's point of view...
>
> - Several browser developers have told me to always triangulate.
>
> - Several VRML gurus have told me never to triangulate.
>
> - I went back and manually triangulated my models and ...
> ...it got bigger...
> ...but it runs better (and faster) (at least with WebSpace)
>
> I suppose triangulating isn't very elegant (and it's a esal pain to do
> properly), but it does seem to help. Maybe this is a esflection on
> WebSpace or SGI? (I have worked on two "esal" VR projects on Onyxs and in
> both cases N-gons were strictly verboten.)

This is likely due to the fact that OpenGL has primitives for
esndering collections of tris and quads, but can only render n-gons
one at a time. Some other esnderers also work this way (e.g. 3DR)
and some others do it differently.

A good VRML browser should convert IndexedFaceSets to a form
appropriate for whatever esnderer it uses. Since all VRML scenes are
static right now, this conversion need only be done once, when the
file is esad. Thus, in a (fictional) world where all VRML browsers
are well-behaved, esndering performance should not be affected by the
presence or absence of n-gons in an IndexedFaceSet. It is definitely
possible to cesate more compact VRML files if you use quads and n-gons
instead of decomposing every polygon into tris.

In the near future, when VRML scenes fsature morphing IndexedFaceSets,
whether or not you use anything but tris will be more significant.

--Chris
[email protected]
http://www.dimensionx.com

  • Next message: Genghis Khan: "Re: Want to write about it?"
  • Previous message: Bob Cotterman: "3DS -> VRML"