[SOLVED] Rigidbody also for child colliders?

Hello Unity Community,

for my Unity 5 2D game I have a character, which consists of multiple components (sprites). For each hand it has separate collider (as trigger). On my parent object of the character I have a rigidbody2d + another large collider (trigger) to detect there is a "danger" near.

As far as I understand (after couple of hours reading through forum threads), I understood that for a collider to work as it should and to not loose performance, a rigidbody should be also attached to it. What I didn't found anywhere is: does that include also child objects?? Do I have to attach also rigidbodies to the hands of my character for colliders to work consistently without performance lost?

Thanks for any response!

 

Thanks for the answer @MelvMay !

.. and to be honest, I expected a different answer, because of something I found a couple of minutes ago: https://docs.unity3d.com/Manual/class-Rigidbody.html

About Compound Collider it's said, that colliders on ChildObjects without rigidbody(rigidbody is on the parent's GO) is the way to go. Am I missing something? I am using Unity 5.5.

Also, the child GOs are not being moved - just the parent object. So, in that case, they are not moving for themselves, but the parent GO (and it's rigidbody) "move them", as the parent object is moving. How to categorize the child colliders and rbs - static, kinematic? Also if I use animation to move the transform of one of the child game object (let's say the hand is moving backwards/forewards), does that mean they are "moving" and I have to attach a rigidbody and set it to kinematic(?) ?

Even more confusing for me - on an enemy (single KINEMATIC rigidbody with a single collider), with a script I am setting in Start() it's velocity (rb2d.velocity) and it moves (gravity and other forces are not applied). As far as I understood, for kinematic RBs the movement should be done ONLY with rb.MovePosition() and setting velocity directly "does not work".. What I am missing/misunderstanding?

 

@MelvMay , sorry, didn't know rb3d and rb2d have different stategy for collision detection. What I am trying to understand, is what the engine expects and does internally, so I can set up my objects accordingly. In my example:

Character (has 1 x dynamic rb2d, 1 x collider to detect collisions with walls ONLY)
--- torso (has attached: 1x sprite, 1 x collider as trigger, which detects bullets that "hit" it, 1x script which handles OnTriggerEnter2d(bullet) ). Not animated, not being moved by a separate script etc.
--- left hand (has attached: 1x spirte, 1x collider as trigger, which detect bullets that "hit" it, 1x script which handles OnTriggerEnter2d(bullet) ) Animated with Mecanim - simple animation to rotate the hand slightly
---- right hand - same as left hand
---- sensor = empty gameobject - with a separate collider, that handles if "fire" or another type of "danger"(but not a bullet) is near character

The question is: do I need kinetic rigidbody2d on left hand, right hand, torso, sensor? On my test it is working without rb2ds, but I cannot tell if Unity is making a physics recalculation because of the GOs("body parts") with colliders but missing corresponding rb2d? The direct answer to that question will solve the thread, can you help @MelvMay?

 

Ok, sir, thank you very much for the answer!

Btw. hands are not children of torso in my case. Torso and both hands are children of character - the parent object. Torso in my case would be chest, if that explains it a bit better. You didn't write about the "sensor", that is the empty game object. Do I need a rb2d there? As far as I understand, from what you tought me here - I don't need that, because I don't directly modify the transform of the "sensor". Is that right, @MelvMay ?

 

Perfect, that's it! Thank you for the support, patience and complete answer to all my questions!