Tue, 21 Jun 1994 22:24:34 -0800

As the proposer of VRIF to this group, let me say a few words in its defense.

> VRIF (is this an official name? I think I picked it up from Jerry)
The working title has been "VR Interchange Format", which turns into VRIF,
although I have not seen that used for it, except for me. I havent seen
much discussion of it (no mail list, etc) from the parties professing
interest originally.

A major point to know is the underlying aim of VRIF, which is similar but
not identical with VRML.
That is, VRIF is for exchanging geometry and world descriptions between
the various commercial, freeware, etc VR programs. There are a growing
number of these, and some of them would undoubtably used for building a VR
based net browser. However, a VR WWW interface needs more information, and
possibly impose some Universal Rules.

Primarily these exist for the PC. By far the major market for VR programs
(in installed systems) is for these VR packages. Comparatively, Mac and
UNIX based VR systems are very rare.

> It assumes that file names are DOS-style.
PC is the largest VR market place, in numbers of installed and potential
sites. That wont change for a LONG while. Name munging dos incompatible
names is a real pain.

>One of the biggest flaws is that it doesn't appear to have any concept
>of type hierarchies. It allows you to associate objects hierarchically,
>but only in the instances (the "objects"), not in the types (the "shapes").
There are two VERY different kinds of objects and object hierarchies in
disucssion here. VRIF is only interested in geometric hierarchies of the
sort normally found in VR and 3D animations. Conceptual (type) hierarchies
comes out of OODesign. Dont make the mistake of confusing them.

>. I'm not sure that their sound mechanism is robust enough
>yet, though (it certainly doesn't have any way of triggering sounds
>dynamically yet).
VRIF has no dynamics in it what so ever. It is a static description of
geometries+rendering attributes. Dynamics introduces the whole issue of
action scripting. Every current VR program addresses this differently.
While there is agreement on a basic set of geometry+render attributes,
there is very little concensus on base sets of dynamic characteristics.

>So far, MSDL is still my favorite, with OOGL a close second (subject
>to change after I play with Geomview a bit)...
Have to dig into these a bunch more. One problem is the higher
complexity geometries allowed - curves, etc. There isnt a decent VR
program I've seen that handles these in real time, 30FPS.

Jerry Isdale
Illusion Engineering, Inc.
2660 Townsgate Rd., Suite 530
Westlake Village, CA 91360

Phone: (805) 371-4530
FAX: (805) 371-4533