There are a lot of things that would be cool. But this is clearly far beyond being reasonable. This would take a massive effort which is useful in very, very few cases. It would make absolutely no sense at this point to waste development time on something like this.
Edit: Everything that has been discussed regarding Bullet is already existing or partly existing functionality. This is something that is at least reasonable.
basically we can combine a fire sim, fluid sim, rigid body sim etc by using obj origins and local points and a grid scale then using the spatial tree to decide what is close enough to what and if it interacts with beyond itâs own processing.
i will try and make a demo using bge when I get a bit and kdtree
someone else with more thinky meat can take over this.
whom interacts with whom (in the local grids overlap)
This does not exist as an open source implementation that is production ready. It would need to be implemented pretty much from scratch, meaning years of development time. Thatâs why it is totally worthless to discuss it and thatâs why this is my last comment about it.
looks like 2.8 today fixes the problem to some extent but cant really test it since rigid body world panel doesnt exist xD have to do all the commands with python
upbge has bullet intergrated - and we can also effect models / save it during a session
(upbge 0.3.0 âwearsâ bpy meshes instead of KX_MeshProxy() )
would be nice if someone were to wrap BT_MultiBody() though