FEA for entertainment

[ivandewolf]

I'm starting an open-source project developing a FEA solver for visual effects work.
I'm attempting to implement this:

http://graphics.cs.yale.edu/julie/pubs/Deform.pdf

I don't know how to compute or store the stiffness matrix. Any tips, hints, or pointers?

 

[40818]

Didn't "Tron" (the movie) use Nastran?

 

[ivandewolf]

Tron was mostly done on symbolics machines, and was mostly rigid forms translating around based on hand-keyframed channels- I doubt if they used any FEA for anything in it....
was "Nastran" around in 1981?

 

[GBor]

The original Nastran code was released to the general populace from NASA around 1971, if I recall correctly.

 

[ivandewolf]

huh. Still, I an't imagine what they would have used FEA for on it.

Anyways, any ideas on how to compute or store the stiffness matrix?

 

[pja]

The first two references in the paper you cite are good places to start...

 

[ivandewolf]

Bathe 1982 and Cook 1981? thanks! I'll try and get a copy of 'em both....
amazon.

 

[GregLocock]

I don't quite see what you are going ot get out of this - linear FEA produces some very odd looking deflected shapes.

There's an old FEA puzzle/game called bridgebuilder (from memory) that may give you some pointers.

Cheers

Greg Locock

SIGlease see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[hondaknight1]

I can see the point I thinkl this is a very interesting idea. Its good for games. I've sadi this before - Greg you're getting old...

I remember a version of Colin Mcrae Rally which was developed in conjunction with PTC who did all the FEA for realistic damage modelling, from which they developed algorithms which ran as the game was played to add damage to the vehicle every time it crashed. The game was a roaring success.

 

[ivandewolf]

did you look at the paper I posted? it's all about a hack that mostly fixes the odd-looking deflected shapes.
Not accurate at all, but, basicly, you rotate your frame of reference on a per-vertex basis.

http://graphics.cs.yale.edu/julie/pubs/Deform.pdf

I just have to figure out how to compute and store the stiffness matrix. at this point, I have no idea. I'd venture a guess that it's stored as a 3X3 matrix, one at every vertex of the tet-mesh, but I can't find a simple answer anywhere. hopefully the 2 books will help.