So since almost all walkable architecture can not be automatically used with collision, i have to use this feature -use complex collision as simple collision-. Could someone outline how much that makes a difference in regards to system resources? Is ok to use this in a level from time to time, is this a non issue on todays standard desktop computer, and only affects mobile devices? Does it make a difference if i use this feature on a static object vs a moveable object?
what is probably still true though - a single vertex in a non-standard collision primitive like auto-convex or using the mesh for collision is as expensive as a collision box. if it will end up having a perceivable impact on performance is hard to say and depends largely on your target platform and mesh complexity, but you should definitely keep that in mind when it comes to the polish stage and you end up having performance issues.