- #1
jorj
- 1
- 0
hello, I am an animation student at bournemouth university in England, I am writing a material simulator for my final major project ( some preliminary results http://www.youtube.com/user/jorjpimm ). The final idea is to have elastic, plastic and tearing and breaking material characteristics, all generated automatically, from a triangular mesh. A lot like http://www.pixeluxentertainment.com/ and their DMM, but mine uses no FEA, just mass and springs.
But I have two problems, one i am currently still using a forward euler integrator, with stiff springs and a tiny time step, any help in a better integrator choice would be appreciated ( i hear a lot about RK4 in games, and my tutors are pushing backward euler.
My other problems is that with a large simulation and collisions and unpredictable things like this, how does one use an implicit integration solution, for example in RK4 i need several samples of f(x) at different points inside h, how do i solve for these timesteps without using an explicit integrator, and then going back and using an implicit one ( which surely undoes any good RK4 might do )
Any help would be greatly appreciated, thanks
- Jorj
But I have two problems, one i am currently still using a forward euler integrator, with stiff springs and a tiny time step, any help in a better integrator choice would be appreciated ( i hear a lot about RK4 in games, and my tutors are pushing backward euler.
My other problems is that with a large simulation and collisions and unpredictable things like this, how does one use an implicit integration solution, for example in RK4 i need several samples of f(x) at different points inside h, how do i solve for these timesteps without using an explicit integrator, and then going back and using an implicit one ( which surely undoes any good RK4 might do )
Any help would be greatly appreciated, thanks
- Jorj
Last edited by a moderator: