Cloth simulation

CoBB · Post by **CoBB** » Thu 30 Dec, 2004 4:56 pm

I made this one from the previous rope program, check it out:

http://www.hszk.bme.hu/~pg429/rongyocska/

Needs glut, windows users can get it here if the program doesn't work (just put glut32.dll in the same dir):

http://www.xmission.com/~nate/glut.html

You should be able to compile it under Debian after installing the freeglut3-dev package.

Post by **Kozak** » Thu 30 Dec, 2004 9:12 pm

It looks cool, what can I say..keep up the good work!

Duck · Post by **Duck** » Thu 30 Dec, 2004 11:04 pm

Looks nice! What strikes me is that the cloth has thickness. I havent seen that in other cloth simulations before. Also, the cloth seems to have hit detection on itself?

Post by **tr1p1ea** » Fri 31 Dec, 2004 2:28 am

WOW CoBB ... THATS AWESOME!

And fast too. Do you think you could stop being a genius for just a few seconds?

.

CoBB · Post by **CoBB** » Fri 31 Dec, 2004 7:51 am

Yes, that's what I'm quite happy with: it has quite fast self-collision detection and cloths can collide with each other too (just release the red one on top of the blue to see it pour down slowly after being shot with a ball). Actually, that's almost the same thing. It also performs surface subdivision on the fly, quadrupling the resolution of the cloth at almost zero cost. Stability is fine as well, I could double the time step at this point without the simulation exploding.

The code is really a big blob of mess, but at least it's fast.

In fact, this was made as an assignment for my Computer Graphics class.

coelurus · Post by **coelurus** » Fri 31 Dec, 2004 11:02 am

Nice stuff, I liked the thing with thick cloths too

EDIT: Ah, I thought it looked weird when I disabled subdivision... I thought the answer would eventually pop up so I didn't dare to ask.

CoBB · Post by **CoBB** » Fri 31 Dec, 2004 1:00 pm

Actually, the thickness is just visual, with the sole aim of hiding the inaccuracy of collisions.

The real simulated system is revealed by disabling the subdivision.