I have a game where I need to detect a bunch of collisions on multiple sides of an object. Because of a bunch of factors, it doesn't really work to check against a single collider and then check relative positions (long story -- due to my particular game mechanics). Best solution I've found is to have several small child objects on top, bottom, right, left sides of the main object, and add colliders to those objects.
Then I can be sure where the collisions are coming from (ie which side of my object is being collided with).
Since these colliders do not have rigidbodies, they are static colliders. However, the parent object does have a rigidbody and it moves around in the game.
Moving static colliders is supposedly bad for performance. But what about parented movement?
Does parented movement of the child objects get penalized even though they are just moving due to parent movement? The solution would be to add kinematic rigidbodies to all of these child objects-- however-- I read here:
That one should never parent one rigidbody to another rigidbody.
Thoughts on best practices? I'm developing for iphone so performance is key.
With my current setup I'm getting significant profiler slowdowns from physics.