Question Can you make a collider ignore force FROM another?

What I want (see pic): Dynamic RBs (red blocks) can ride on either kinematic RBs (moving blue platforms) or other red blocks, and still take forces from static RBs (green terrain).

Moving/parenting transforms do not work. Force override for whole layer and IgnoreCollision aren’t specific enough.

Question: Is there a way to either 1) Make collider A not receice force from collider B (if both have same layer)? OR 2) Make a collider ignore incoming forces from a specific direction that involve colliders of a given layer(s)?

If I could do either, then each block would respecte terrain, support blocks above it (send force), and not affect trajectory of blocks below. Assume my code can figure out what collider/obj is “riding” another. (Same problem as another thread, but this is a different aspect of that problem)

 

Thanks so much for taking time to reply.

To clarify: Since force overrides take in a layer mask, if the block in the middle stops receiving forces from the block layer, then it’s not affected by the top block (good), but falls through the bottom block (bad). Send force overrides and Physics2D.IgnoreCollision() have the same issue. I want to block force recieve (or send) between two specific colliders, without influencing how it interacts with the whole layer.

Bummer. Another clarifying question, given I can’t get into the physics simulation step or modify contacts. You mean there’s no way to tell the physics simulation (before it runs) to request it ignore all force from colliders A to B but include B to A?

Alternatively, do you have any suggestion to achieve this desired behaviour? I’ve been going mad trying to get one dynamic RB to ride another without affecting it, AND allowing the rider to move normally with AddForce in the reference frame of the thing it is on.

 

Thanks. This helps a lot, as it’s so hard to know what doesn’t exist.

I’ve been trying to break up collision callbacks, sending forces, and receiving forces. Unfortunately, I think the workaround is to give each block two duplicate colliders. One has collisionCalbacks/recieve forces, and a duplicate one with only send forces. This way I can tell Physics2D to ignore collisions between blockA forceSendCollider and blockB forceReceiveCollider. This way B can ride A without affecting A’s trajectory. Sounds expensive, but from this convo, it sounds necessary.

 

My thinking is that IgnoreCollision is a function of two colliders, not rigidbodies. If I duplicate a collider into one collider that only receives force (send force layermask = 0), and one collider that only sends force (receive force layermask = 0) on the same rigidbody, then I can call Physics2D.IgnoreCollision(blockARecCol, blockBSendCol, true).

This way, for blockB riding blockA, if blockA moves around (flying, riding, etc), it moves as though blockB isn’t there (Physics solver ignores blockArecCol-blockBsendCol, so B can’t send force to A’s rigidbody), blockA still sends forces to blockB (blockBrecCol-blockAsendCol collisions not ignored), and they still send callbacks (blockAsendCol-blockBsendCol not ignored). Repeating for blocks B/C/D would let me make stacks where force only transmits to the block riding, and not down. Then B can ride A by friction (+ rigidBodyB.AddForce to maintain contact with A), and all these extra forces don’t affect A’s trajectory (because it ignores forces from B). That’s the sort of behaviour I’m aiming for.

 

Isn’t that right then? I might simplify by making blockArecCollider ignore both block B colliders. Whenever A and B collide:

• BlockASendCollider - BlockBSendCollider: Makes contacts, sends collision callbacks. Both have recieve force layer = 0 => no forces.
  
• BlockASendCollider - BlockBRecCollider: Makes contacts, send + recieve layers match up for contacts to generate force on B, but no forces on A.
• BlockARecCollider - BlockBSendCollider: Ignored => no contacts => no forces
• BlockARecCollider - BlockBRecCollider: Ignored => no contacts => no forces

Net result: A sends force to B, and both get collision callbacks. And all of this doesn’t impact ignoring/excluding collisions from other RBs in the same layer.

To be clear, I’m proposing giving each RB two colliders of the exact same shape, size, and offset.

Is there something I’m missing here?

 

Update: It kind of works. If you make a duplicate collider, and set it up like I explained, it will allow you to send (and not receive) force from another specific RB.

I do need to abandon it since I have no way to go into the physics system to make the edits I need to apply proper forces for the blocks above to ride during the physics sim step. My next plan is to make everything kinematic, and effectively code all my dynamic RBs to work with kinematic RBs that calculate next position in sequence. This way I can mostly avoid the physics simulation doing any of the real work, since I clearly don’t have enough access/control via the normal API.

If only I had a Joint2D that would only apply force to one of the two objects. Or the ability to modify contacts during physics simulation. Without either, I need to do my physics in a fundamentally different way.