
I just looked through the samples and noticed that everything is decomposed to be convex. So is farseer going to be going back to just convex polygon collisions? Because the distance grid did work well for the complex shapes. For a shape with lots of concave
features does the convex decomposition perform faster than a grid based version? I am also working on coupling this to a fluid simulator so there will be a lot of particlebody collision checks.



FPE 3.0 is based on Box2D that uses the SAT algorithm for the narrow phase. I've not tested the SAT algorithm versus the distance grid, but I think that the SAT algorithm is faster (when testing with convex polygons only). The decomposing can be done before
the game starts, just like the precomputation of distance grids.



Are there any plans to include the distance grid stuff in FPE 3? Also I'm interested in developing some fluid/soft body dynamics for FPE (http://youtube.com/kotsoft). Could I join the project, or is Farseer strictly
rigid body physics?



I don't have any plans on including the distance grid in future versions of FPE. About the project membership, I will contact you by mail in just a second.

