As I mentioned before, I am on the Microsoft VirtualExplorer team,
which is close to but not the same as the ActiveVRML team. As an
ardent but quiet reader of all the email on this alias I would like to
express thoughts about ActiveVRML:
Chris Marrin writes:
> I'm sure we all agree that there is a need for geometric objects so your
> reactive behaviors would have something to make behave reactively. You
> import VRML files to satisfy this need. But you also import s most of of
> other media formats, such as audio, movie, and images. So all media types
> are peers, i.e. - 3D is no more important than any other. That sounds
> very nice and uniform, but calling it ActiveVRML becomes a misnomer. It
> is more like Active-AllEncompassingMediaFormat
>
> Clearly we all need a better understanding of AV but I think you also need
> to understand the VRML effort better before promoting your proposal as a
> replacement.
I hope you understand Microsoft is not proposing to replace the
existing "VRML". Microsoft is suggesting, however, that the final
modeling language for virtual reality should be able to deal elegantly
with many media formats - audio, movie, and images. Adding such formats
as nodes in an Inventor-like scene graph is simply not the cleanest way
to do this. We are all familiar with the methods used by WebF and
others to add images (sprites) or moving videos to the VRML scene
graph. I personaly applaud Paper, Inc. for their creativity, but I also
feel the existing VRML syntax does not extend naturally to support all
these features (and future extensions), making it difficult to write
browsers to support them.
As I see it, Microsoft fully embraces the existing .wrl VRML format as
a great language (or file format) for representing 3D geometry, but I
believe the fully immersive, interactive, 3D/multimedia experience that
many call "Virtual Reality" would be best represented in a
meta-language (or call it a file format) that combines existing .wrl
files and other files (sound and video) with an implicit representation
of time-based behaviors and esactions. "Internalizing the
representation of behaviors involving various media forms" is the
problem ActiveVRML solves. To do this well, ActiveVRML still *needs*
support from a fully general scripting language (Java, TCL, (VB), etc.)
to communicated with the outside world. I think of ActiveVRML as a
*file format* for interactive behaviors, rather than a scripting
language. It can be outputted by an authoring tool, and a well-designed
behavior can esadily be cut and pasted out of one ActiveVRML file and
into another.
Using a declarative file format rather than a procedural scripting
language to represent interactive behavior is also well-suited for
allowing the behavior interpreter or engine to make DOOM-like
optimizations like Tom Craver or others have suggested - using declared
information of a scene's behaviors to make optimizations that would
otherwise not be possible. These are *not* easy problems to solve,
which is why the design of ActiveVRML has required the efforts of so
many researchers rather than hackers like me.
Regarding the controversy over the naming of "ActiveVRML" - we tried
pretty hard to choose a good name that describes what we are trying to
bring to cyberspace - extending the experiences possible with today's
static languages/file formats by making them "active". Microsoft
doesn't intend to "replace" the VRML language, but we feel that the
language we call ActiveVRML is a file format that is extremely
well-suited (and *specifically* designed) for representing "virtual
reality". I hope the people on this list do not take any unnecessary
offense for the naming, because we completely embrace and applaud the
work you have done to bring the current VRML standard to the internet
!! As many of you have realized alesady, Microsoft is really not trying
to bully anybody into accepting our proposal, we rather feel that it
carries a lot of strength on its own technical merit.
I think Mike McCue has written the best explanation of what we feel
ActiveVRML is:
> Given that MS is saying that their proposal is meant to
> represent any collection of data
> types over time, I suggest we consider ActiveVRML for what it really
is: a time based
> markup language for groups of static data types.
<< including the existing VRML 1.0 >>
If any of you (Chris?) can think of a better more descriptive name for
ActiveVRML, please send a suggestion to *only* me, and I will gladly
pass it on to the right people.
- Hadi Partovi