Page 1 of 1

Cloth simulation

Posted: Wed Apr 20, 2011 12:05 am
by saggita
I made a cloth simulation using c++. It uses implicit backward euler integration and continuous collision detection for cloth-object and cloth self-collision. Strain limiting was used to simulate inextensible cloth. BVH/AABB was implemented to avoid n^2 collision checking. Rendering was done in Maya using AnimOBJ(http://code.google.com/p/animobj) which is made by myself as well.

Below is a link to the short video.

http://www.youtube.com/watch?v=YVhD2d2wWHU

I am currently trying to implement global collision treatment methods based on papers below.
If you have some experience with global collision treatment, please share your idea with me. I implemented the second one (volume preserving impulses) and have some problem with convergence. I believe the matrix is not really definite and need to use GMRES. I am using conjugate gradient now.

Also the first paper is somehow confusing.

Thank you.

Re: Cloth simulation

Posted: Thu Jun 27, 2013 5:40 am
by saggita
Since my previous posting, I've been developing my own simultaneous proximity and continuous collision handling method for cloth simulation. I tried a couple of approaches like ones linked above and was not satisfied much. Below is a link to my current approach in youtube.

http://youtu.be/6yKrcV16k6A

The idea is basically creating a big linear equation and solve it rather than treating individual collisions iteratively until it converges. It is kind of similar to LCP but I am using impulses and proximity information in position level rather than velocity level. The matrix of linear equation can be over-constrained and ill-conditioned due to the complexity and irregularity of collision distribution. I overcome the problem by using conditional regularization. This research is still on-going but I just want to share it.

Attaching a screenshot to show the stress test with more clothes.