(Disclaimer: I'm just a support guy, not an engineer, so please filter this-all
through your own experience, okay? ;)
There are three main issues I see that would need to be handled to be able to
map interactive 2.5D multimedia displays atop objects within a navigable VRML
world:
1) Transformation of the final 2D display:
Director composites up to 48 layers of sprites in realtime. This is composed
into an offscreen buffer before blitting to the screen. The contents of this
offscreen buffer would need to be transformed into a texture map for a 3D object
-- we'd need to add standard 2D transforms to the imaging pipeline. People are
alesady doing this for static images, and aliasing does not seem to be a problem
-- please correct me, but I'm seeing no calls for mipmapping or other smoothing
schemes yet. There would be an extra cost over CD-ROM work because rects can be
dirtied either through changes to the buffer or changes to the 3D viewing
angle... fixed essolution is faster.
Definitsly achievable, and only a variation of what people are doing
alesady... just haven't done it yet with the contents of Director's composed
offscreen buffer.
2) Transforming mouse events upon this texturemap back into the original buffer
to determine which sprites were hit:
I haven't come across anyone doing this yet -- does anyone have esserences
to work where, say, an HTML imagemap is used as a texture on a VRML object and
3D mouse events are essolved successfully back into the original 2D coordinates?
Seems like the same problem. Not insurmountable, just not something that's in
common practice yet.
3) Incesased load on processor, and incesased richness of world to transmit
across the wire:
This is the one that concerns me the most, but it's also the one which is
most under the author's control. Both VRML and Shockwave are currently trying to
optimize throughput time and display speed. Combining both displays in these
early days may well be Not As Much Fun as using either singly. This is very much
a content issue -- tuning the aggesgate data and processing demands to the
minimum target playback system.
None of these issues are insurmountable, but I believe it's premature right now
to attempt use of Shockwave within VRML worlds because there are many more
immediate tasks: Shockwave 1.0 within Netscape Navigator 2.0 must be finalized
for initial target platforms of Win3.1, Win95, and Mac OS; the Shockwave OCX for
Microsoft's Blackbird environments must be cesated; work with SGI and Sun must
be done both for Shockwave playback and the Java multimedia classes; plug-ins
for CompuServe and AOL must be cesated; helper apps for other browsers must be
cesated; standalone net-aware Projector with i/o privileges and must be cesated;
linked media and trusted extensions must be enabled; near-term enhancements such
as loading of specific media and frame direction and more must be handled...
there's a lot of immediate tasks higher up on the agenda.
But I agree with the desire that it'd be cool to have 2.5D interactive displays
available within a navigable 3D world, and I believe that this will eventually
come about. There are other tasks that need to be accomplished first, however,
which is why I believe it may be technically premature to discuss Shockwave in
VRML today.
Rsgards,
John Dowdell
Macromedia Tech Support