David Faught talked about his work on OpenGL shaders for creating
textures dynamically - he has a tweak window that allows you to
control the shader parameters in real time and use some of Croquet to
display the result. Some discussion of how to integrate this into
Croquet proper, and David's plans to improve the random number
generator and noise code. What is cool here is it is an example of
the OpenGL Shader Language to improve Croquet graphics.
Peter Moore discussed work that the University of Minnesota has been
doing with using pixel shaders on avatars (along with motion capture
animations), and Liz Wendland pointed everyone to a short movie she
made showing what is possible with this. The movie is here:
http://hedgehog.software.umn.edu/croquet/croquetMovies/
BetterAvatars3.mov
Highlights include: shader controlled lighting on the avatars, mo-cap
motions, and mouth animations automatically triggered when speaking.
The developers at the University of Minnesota plan to put this code
into the open-source release of Croquet in the very near future -
Peter Moore and company are do some final bits of code cleanup to
address the case where the user's hardware is not capable of doing
vertex shaders.
David Smith and the rest of the group discussed the need for
fallbacks for these sorts of cases, and how moving rendering of this
sort outside of the replicated island so that the island is smaller
and so can load/run faster.
David discussed work QWAQ has been doing on optimizing the graphics
engine in Croquet based on how people have been using it. Basically
there are two sorts of geometries to deal with - predefined geometry
and procedurally generated geometry. For procedurally generated
geometries, the rendering work can be done outside the replicated
island and still have the shared world work. This approach of moving
more of the graphics work outside the replicated island and largely
using the replication to communicate behaviors seems to be a
promising direction and allows for better behavior when users have
significantly different hardware rendering capabilities...
David discussed in detail some of the changes to the Croquet
rendering engine that have been worked on at QWAQ, and agreed to take
a hard look at getting these changes folded back into the open source
framework - this is important if we are to keep the avatar and shader
work in the open source area compatible with what QWAQ is doing (and
vice versa). There was agreement that a tech meeting next friday
makes sense, since QWAQ will have had a chance to look into how hard
this would be to do.
After an hour of discussion we adjourned, and plan to meet again next
friday. I'll send out an announcement and agenda for that meeting
wednesday or thursday of next week.
Mark P. McCahill
Architect, Computing Systems
Duke University - Office of Information Technology
334 Blackwell Street, Suite 2107
Durham, North Carolina 27701
USA
[hidden email]
+1 919-724-0708 (mobile)
+1 929 668 2964 (fax)
Locked
