pretty cool, also keep in mind that Bullet has 2D support as well, which we will get for free once we move to Native Bullet completely. Whether it's close in performance to a full 2D engine, I'm not sure
According what I have read so far, using a 3D physics engine for 2d Physics calculations is a bad idea for performance point of view (and it sounds logical). I’m not familiar with Bullet, so I don’t know if Bullet has some optimizations for 2D calculations.
Nice ! i'd really like to see these things on JME engine like also @t0neg0d are doing . I didn't know about the existence of Dyn4j but as @erlend_sh said the fact that is written pure in java is a good point . Did you know about liquidfun , yes is a box2d fork but the particle and softbodies features are really cool! :)
I didn't know about liquidfun, thanks. I will take a look. jBox2d and Dyn4j are for simulation rigid bodies. I prefer to star with one of those two first making a very basic integration (an appstate, a class to handle the physics world interactions with mesh and a few controlers). Keep it simple let say, then add more funtionality.
I appreciate all your suggestions and ideas. Thanks