I'm trying to think of the best way to design a giant 2d destructible object made of squares. Each square having its own collider so you can destroy pieces of it and disconnected "chunks" will fall off. My first thought is to have a fixed joint for each of the squares neighbors, this way as the squares are destroyed the chunks should fall off automatically. I'm just a bit worried that having massive amounts of joints will have performance implications. Another way could be to have all the squares as children of a parent object, and every time one of the squares is destroyed, I could check to see if any squares are "disconnected" and parent them to another object - but that seems pretty heavy for runtime code. Can anyone think of some better ways of doing this, or have your own experiences of what works and what doesn't?