Adding thickness to geometry for bullet sim

[soomld]

Hi all,

 

I have been learning to use bullet sim for the past few weeks and was trying to add thickness for my breaking teapot. I have unchecked "Create Inside Surface" at the Voronoi fracture node to remove the insides, but currently the surfaces are paper thin. I have tried all the extrude nodes, but my surface points seem to get wonky.

I am also wondering if there is an effective way to auto freeze the small pieces from rolling around, I have tried some solutions from the forum but it haven't completely stopped it. I am guessing it is a drawback for using bullet sim or I shouldn't be cracking the geometry into really small pieces.

 

 

I have attached a picture and file too

Thank you!

voronoi_v002.hipnc

[davpe]

Hi,

 

regarding first part of your question, you should be able to add thickness to the geo easily with polyExtrude node after converting teapot to polygons and before fracturing it. Then fracture it and use it as an input for Bullet solver.  

 

secondly, I don't think you can do freezing of moving pieces without getting really weird results. There is something called Sleeping in Bullet solver that deactivates those pieces that cross certain threshold. That will solve some flickering, but not wildly rolling pieces. Rolling is mostly caused by converting your geo into convex hulls for collisions which tend to have roundish shapes (you can display them in Bullet solver node). Thing is that Bullet is designed more for speed than accuracy. That is why it is so good for making large sims where inaccuracies like this are not too apparent. If you need fine looking movement you should probably go with brute force RBD solver which is much slower, eats up more memory, but is also much more precise. You will find however that its setup differs quite a lot from Bullet and it is not as convenient for complex setups.     

 

Also, having a good geometry makes a big difference in how stable Bullet is. Using messy and non manifold (closed) meshes results in wrong calculations and overaly worse looking (or terribly looking possibly) simulations. So using paper thin geo that have no volume after shattering is causing at least part of your problems.

Edited August 26, 2015 by davpe