[Bf-committers] Deformable model source code + demo

David Bourguignon david-m.bourguignon at wanadoo.fr
Fri Sep 24 19:50:12 CEST 2004


Hi Herman,

Concerning the "mesh with dynamic parameters" data structure, the 4 experiments
you can run with my demo package show clearly that the traditional mass-spring
system (masses located at vertices and springs along edges) is not a satisfactory
model for volume meshes. However, it is well known that it does a pretty good job
for surface meshes (a good approximation for clothes and other thin "sheets").

The new deformable model proposed allows to simulate the mechanical behavior
of the material without explicitly using the volume mesh geometry for force
computations, as opposed to m-s systems where force directions are always
along edges. In that respect, the new model is closer to a finite element approach,
without the problems encountered with elements during large deformations.

Finally, it could be interesting to try using this new deformable model for cloth,
using triangle elements instead of tetrahedra. (See the conclusion and future work
section of the paper.)

Cheers,

David.

> From: Herman Bruyninckx <Herman.Bruyninckx at mech.kuleuven.ac.be>
> Subject: Re: [Bf-committers] Deformable model source code + demo
> To: bf-blender developers <bf-committers at projects.blender.org>
> 
> Nice!
> 
> Does anybody have already an idea what (if ever) should be changed/added to
> the data structures in Blender that represent objects, scenes, etc? I can
> imagine a new kind of object is needed: a "mesh with dynamic parameters"
> (i.e., each node has mass, and the arcs between the nodes have elasticity
> and damping).
> 
> I have similar "design questions" when trying to understand how best to
> integrate "robots" into Blender: robot joints also have damping and
> elasticity, and links have mass (and possibly elasticity and damping
> too)...
> 
> Herman

-- 
http://perso.wanadoo.fr/david-m.bourguignon/



More information about the Bf-committers mailing list