From: Robert Zeleznik [bcz@cs.brown.edu]
Sent: Sunday, October 31, 1999 8:32 AM
To: 3d-ui@hitl.washington.edu
Subject: VR/desktop graphics environments


I'd appreciate it if anyone can respond to this unfortunately long
email of mine before this coming Friday.

Thanks,
-Bob



Before making a proposal to our sponsor, we (my group) wanted to see if 
there would be interest in the community for a broad scope, open-source, 
cross-platform graphics system.  The goal of this project is to bring some 
synergy to the computer graphics community so that we can share, replicate, 
and extend each other's work.  In doing this project, we are not trying to 
reinvent the wheel -- in fact we expect that much of the code for this 
system already exists and only needs to componentized, refined, and 
documented.  However, before engaging this project we very much need feedback 
from the community to determine the relevance and importance of this project.  

Nearly everyone in the graphics research community has their own unique 
"system", which has made it particularly difficult for the community to 
benefit from the synergy of shared results and scientific replication.  
Attempts to provide standardized systems typically fail because of issues 
related to intellectual property, non-portable software, monolithic design, 
and limited functionality.

Thus, our proposal is to consolidate a wealth of systems design into a 
single open source, cross-platform (NT/IRIX/Solaris/AIX/HPUX), C++ component 
archictecture.  Programmers could choose to use only the components that 
they needed without the bloat of the full system -- ranging from just 
sharing algorithms/data structures components like a subdivision surface or 
gesture recognizer all the way up to sharing the complete application 
infrastructure.  As an open source system, people could extend the system 
by adding new components, or by re-coding existing components.

The system's functionality will be based on a coarse-grained set
of low-level modules and a potentially finer-grained set of higher-level,
perhaps application-specific, modules.  The intended functionality of 
this system, which will be derived from a variety of existing code 
bases (perhaps yours), includes:

       portability layer: (probably Netscape's portability layer, NSPR)

       component infrastructure: (probably Bamboo from NPS)

       display configuration: (standard desktop, workbench, cave, etc)

       devices: (mouse, tablet, data glove, polhemus).

       scene management: (high-level objects and/or scene graph nodes with
                          access to OpenGL and/or Performer-equivalent)

       encapsulated interaction techniques:  (FSAs, Interactors, 
                          two-handed techniques, immersive/desktop, etc)
                        
       geometry components: (subdivision surfaces, CSG, NURBS?)

       sci-vis modules: (regular/irregular grids, streamlines, etc) 

       rendering: (conventional and non-photorealistic rendering)

       web documentation: (similar to Java doc plus examples)


******** QUESTIONS ********
We really need feedback from the community including general reactions
and specific replies to the following questions:

If delivered as promised, would you *actually* use this system? 

How would you use the system?  Why?
   (i.e., to see other people's stuff?
          as your primary development platform?
          just for some components like devices and display configuration?)

What technical concerns would you have for this system?
   (i.e., must be able to handle billion-polyon models or else...,
          must have a Java/Python/etc scripting interface)


What political/social concerns would you have?
   (i.e, "I don't trust Open Source", 
         "I've got too much invested in my own system",
         "Only if there were good authoring tools",
         "My stuff will be better than this"",
         "Nice idea, but I'm too busy to switch to it")

Would you have interest/time incorporating your own work on whole
   or in part to the system?