Restitution and friction per collision shape

[dans]

Ive written code to add per shape restitution and friction, though that I would sumbit for consideration to add to the main trunk (Would save us some merging in the future . I have attached the effected files along with this post (did this work with 2.71)

It does change the API in that btRigidBody::m_friction and m_restitution now become obsolete and are hence those and their accessors have been removed.

Cheers

[mickey]

hmm, just wondering, why would the restitution and friction be on the shapes and removed from the bodies?

[Erwin Coumans]

It would be good to have per-shape friction and restitution materials, but it seems we only need this for btCompoundShape isn't it?

For single shapes, we just have the current per-body material settings, and for triangle meshes there is already btMultimaterialTriangleMeshShape.h.

So a new btMultimaterialCompoundShape is preferred. Do you consider implementing this, or shall we add it as a feature request in Google Issue tracker?
Thanks a lot,
Erwin

[dans]

Mickey, Ultimately I did it this way to avoid the check for btCompoundShape when calculating combinedRestitution/Friction. This format handles every combination compound or single shape without performing that check. This makes btRigidBody::setFriction/Restitution unnecessary. To maintain btRigidBody::setFriction etc one could iterate all shapes and set shape params from input values, but this would be fairly pointless in using multimaterial compound shapes as the entire set ends up with the same parameters. Erwin does addresses this with btMultimaterialCompoundShape.

Erwin, As it is the preferred method I am happy to implement btMultimaterialCompoundShape (has to be on my own time not my works ) I will submit for review when it is done.

[mickey]

Hi Dans

Thanks for the info!

[natepak]

Has progress been made on this front? Collision shapes with friction and restitution would be very useful.