Re: Perfomer Morph Node

New Message Reply Date view Thread view Subject view Author view

Jim Helman (jimh++at++surreal)
Mon, 19 Sep 94 21:04:44 -0700

--------
The SIGGRAPH paper describes what we've done, which is (usually)
more than what we've released. pfMorph won't go out until the
next major release. At this point, specifics aren't that useful,
since we'll be chewing on it a bit more before spitting it out.
But here's an outline.

Rumors that we're all fans of the schlockiest TV show to disgrace
the Sat AM airwaves and that Performer 2.0 will be named the
Mighty Morphing Performer Release are utterly false. We're
morphing because it's important for 1) morph LOD and 2) real-time
character animation. Morph LOD has visual advantages over
hard-switching and performance advantages over fading (fade LOD
has some bad characterstics since stress-based LOD scaling may
draw both models, thereby increasing geometric load before
decreasing it.) Real-time character animation is useful in
entertainment applications (like Disney Imagineering's Aladdin VR
ride), and in digital production where it permits
performance-driven animation and recording real-time to video
rather than frame-by-frame.

The basic capabilities for geometric morphing are: 1) weighted
(linear) combination of pfGeoSet attribute arrays (e.g.
coords, normals, colors, texcoords) as inputs to generate a
renderable output for each attribute. 2) multiprocessing of
the morph computation across host CPUs. 3) multibuffering of
the output arrays to support the multiprocessing. So morphing
isn't a standalone feature, it also overlaps with dynamic
multibuffered geometry and with LOD.

You could implement these features on top of Performer right
now, or wait. But in either case, if you're modeling new
databases, you might want to construct extra LOD models with
the vertex correspondence required for morph LOD, e.g. when
modelling a house with 2 LOD's you need an extra intermediate
model for the morphing:

          /\
         / \
        / \
       / \ ________ ________
      /\ /| / / /| / /|
     / \ / | / / / | / / |
    / \ / | / / / | / / |
   /______\/ | /___/___/ | /_______/ |
| | | | | | | | |
| | / | | / | | /
| | / | | / | | /
| | / | | / | | /
|______|/ |______|/ |______|/

  high detail intermediate low detail
    model model model

rgds,

-jim helman

jimh++at++surreal.asd.sgi.com
415/390-1151

New Message Reply Date view Thread view Subject view Author view

This archive was generated by hypermail 2.0b2 on Mon Aug 10 1998 - 17:50:33 PDT

This message has been cleansed for anti-spam protection. Replace '++at++' in any mail addresses with the '@' symbol.