[Soc-2013-dev] Weekly Report #9 Rigid Body Simulation Improvements

Sergej Reich sergej.reich at googlemail.com
Sun Aug 18 11:51:49 CEST 2013


Sorry for the late report.

This week I spend most of the time on convex decomposition.
Tried replacing HACD with the V-HACD library. It gives better results
and is faster, however it also has several limitations like not being
able to handle non-manifold meshes.
So at this point I'm thinking of going back to regular HACD and fix the
performance problems myself.

I started to implement a deactivation system.
Right now it's still primitive and can only activate objects on
collision.
We'll also need a way of creating what bullet calls "ghost objects". So
one could activate rigid bodies based on collision with a mesh that
doesn't interact with the simulation. The idea right now it so add a new
rigid body type for that.

Also added continuous collision detection for rigid bodies.
This will have to be a user option I fear as it doesn't work well in all
cases and has an impact on performance.

Next week I'll add different triggers and ghost objects for the
deactivation system and work on smaller TODO items.
I'll take a break from convex decomposition for a wile, it just takes
too much time from the other tasks.




More information about the Soc-2013-dev mailing list