[Bf-committers] IRC Meeting Minutes, 15 Aug, 2004

[David Bourguignon]

> IRC Meeting Minutes, 15 Aug, 2004 
> 
> 2) B-Con 1: possible 2.35 features & roadmap 
> 
> * softbodies (Ton R/Jensen O.M.) 

Hi all, 

I missed last IRC meeting, but in a previous one, I said I was willing to 
help with the softbodies. Since I have been very busy recently (finding a 
job...) I haven't made any progress. According to last meeting minutes, Ton 
and Jensen are in charge now, so I just want to give a recap of what I was 
thinking about. 

I wrote roughly a year ago an overview of softbodies simulation in computer 
graphics. It it is part of an heavy (~30 MB) pdf you will find here: 
http://perso.wanadoo.fr/david-m.bourguignon/pub/Bou03/ 
However, I have put (temporarily) a relevant excerpt here: 
http://perso.wanadoo.fr/david-m.bourguignon/tmp/previous_work.pdf 

IMHO, there are two interesting approaches for softbodies in Blender. In 
both cases, you need to transform your surface triangular mesh into a volume 
tetrahedral mesh by (temporarily) inserting internal vertices. (Not an easy 
problem if you want to obtain "nice" tetrahedra, but I think several solutions are 
possible.) 

* First softbodies approach: discreet model. 
A paper on this method is available here: 
http://perso.wanadoo.fr/david-m.bourguignon/pub/BC00/ 
- Pros: easy to code (extension of mass-spring systems), easy to understand, 
easy to tinker with (IMHO, this is very important for artists). 
- Cons: rough approximation of the real physical behavior, eg, it is 
difficult to obtain quasi-incompressible behavior. 
I have to clean the code written for this paper, compile it under linux and 
mswin to provide a nice demo. I think I can do this before the end of next 
week. 

* Second softbodies approach: continuous model. 
IMHO, the most relevant paper is: 
G. Picinbono, H. Delingette, and N. Ayache. Non-Linear Anisotropic 
Elasticity for Real-Time Surgery Simulation. Graphical Models, 
65(5):305-321, September 2003. 
You can find it here: 
ftp://ftp-sop.inria.fr/epidaure/Publications/Delingette/Picinbono-GM03.pdf 
- Pros: classical continous mechanics equations solved in a very 
efficient way, thus the model is fast and accurate. 
- Cons: difficult to tinker with the model without a good continous 
mechanics background, and thus potentially difficult to use for artists. 

That's all! Sorry for the long post. I welcome any question you may have. 

David. 

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