[Bf-committers] Bullet physics plans/status/update

[Matt Ebb]

On 10/10/2006, at 09:41 AM, bjornmose wrote:

> Sure erwin is doing excellent work, but asking for a 'one physics  
> engine fits all' is like all physisists asking  for GUT (grand  
> unifing theory).
> The search for it is what the best (and lower) paid heads in  
> physics are working for and i guess it needs another bored swiss  
> patent office bureaucrat to be struc by the grand insight before we  
> will see the light and beauty of it...
> And even if, i doubt that any CG software is able to struggle its  
> way throug the maze of 'total antisymmetrische Masstensoren' to get  
> any usable result.
> So for time beeing, i'd suggest we'd better stick to what CG needs:
> Fake it! and fake it the best you can ... even if you need to use
> different tricks (tools) to get there.

I've got no idea what's involved, and I'm sure it's extremely  
difficult stuff, but it does seem possible to the extent that  
Autodesk are working on a unified physics system for a future version  
of Maya. Of course they're not trying to simulate every particle in  
the universe, rather they have designed a system that can be reused  
for different types of interaction. It looks like they have got some  
quite impressive results so far as demonstrated by these pictures of  
a talk given at the Eurographics 2006 conference:

http://www.cg.tuwien.ac.at/~ralf/JStalk/

Perhaps this is totally unrealistic for us since it might mean  
rewriting all the physics stuff at the same time. Though I also  
remember (what I Think Tom M was talking about) that Ton was talking  
at some stage about providing a 'force API' where objects or geometry  
could provide information about what forces they are exerting or  
feeling, so other objects could check that and react accordingly.  
Would something like that not be feasible?

cheers

Matt


------------------------------------------
Matt Ebb . matt at mke3.net . http://mke3.net



-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://projects.blender.org/pipermail/bf-committers/attachments/20061010/2293586b/attachment.htm