P2 Physics - CollisionGroups - Bitmasks

[wayfinder]

Hey there!

 

I'm trying to wrap my head around collisiongroups in p2 at the moment, and I am hitting a wall. I can set up my game without collisiongroups, then my player will collide with all objects (only the ground for now) in the world, but try as i might, i can't get it to collide with the world bounds. OR i can set it up with collisiongroups (one for the player, one for my ground), then the player WILL collide with the world bounds, but not with the ground any more.

 

I have no idea why, the world bounds example is so simple and yet when i try to do something similar, it just doesn't work!

 

I have to confess, I don't *get* CollisionGroups yet. 

 

Is there a document that talks about best practises and the general concepts somewhere? the examples are really quite opaque to me  

 

 

he;lp

[wayfinder]

Does anybody know anything about this? I can't get my stuff to collide ;(

[lewster32]

I believe this may be of help - it looks like collisionGroup is based on Box2D's implementation.

[wayfinder]

yeah, i read that earlier today, but i can't seems to get phaser to agree with me on what should be happening

 

I set up my player to collideWorldBounds, and he doesn't

 

I set up my player to collide with my level collisiongroup, and my level to collide with my player collisiongroup, and they don't

 

This is a little frustrating

[wayfinder]

update: i can let my player collide with the world bounds if i comment out the line that assigns him to a collisiongroup.. maddening

 

 

edit: update 2: if I remove the line that assigns a collisiongroup to my level, a separate sprite that has no collisiongroup set will collide with it. not, however my player? why. argh.

 

 

edit: update 3: if I remove the body.collides on my player, it will collide with the level, but not any more with the bounds. i feel like i'm driving a clown car

[wayfinder]

Another update: by moving the setBoundsToWorld instruction before the sprite collisiongroup assignment, I have now created a situation where my player (assigned a collisiongroup and collideswith set to contain the level) collides with the world bounds (yay!) but not the level (boo!). the level has a collisiongroup assigned and collideswith contains the player collisiongroup.

 

Is there a way to access the CollisionMask directly(other than via collides()?

[Zef]

I don't know if this is still why you ware having problems, but it definitely was to begin with. Take a close look at this straight from the examples  

//  Create our collision groups. One for the player, one for the pandasvar playerCollisionGroup = game.physics.p2.createCollisionGroup();var pandaCollisionGroup = game.physics.p2.createCollisionGroup();//  This part is vital if you want the objects with their own collision groups to still collide with the world bounds//  (which we do) - what this does is adjust the bounds to use its own collision group.game.physics.p2.updateBoundsCollisionGroup();

[wayfinder]

Yeah.. and it turns out my scenery shapes had the wrong collisiongroup set I was treating setCollisionGroup as an attribute instead of a method in that one call... that kept me busy until just now, with impossible-to-extrapolate-from behavior.

 

I'm writing this down as much for my own memory as for maybe anyone else looking for information who happens to stumble on this thread:

 

 collisiongroups:   0 0 0 0  0 0 0 0  0 0 0 0  0 0 1 0  (bounds collisiongroup, decimal: 2)  0 0 0 0  0 0 0 0  0 0 0 0  0 1 0 0  (player collisiongroup, decimal: 4)  0 0 0 0  0 0 0 0  0 0 0 0  1 0 0 0  (enemy collisiongroup, decimal: 8)  0 0 0 0  0 0 0 0  0 0 0 1  0 0 0 0  (scenery collisiongroup, decimal 16)-------------------------------------------- collision masks   0 0 0 0  0 0 0 0  0 0 0 0  0 0 0 0   (zero mask: Will collide with nothing)  0 0 0 0  0 0 0 0  0 0 0 1  1 0 1 0   (player mask: will collide with scenery, enemy and bounds: 16 + 8 + 2 = 26)  0 0 0 0  0 0 0 0  0 0 0 0  1 1 0 0   (bounds mask: will collide with players and enemies: 8 + 4 = 12) 

 

and once again, a stupid bug in my code kept me tinkering for two days

[wayfinder]

To Phaser's credit, its system is a lot easier than working with the raw masks – if you don't accidentally write setCollisionGroup = x instead of setCollisionGroup(x)

[wayfinder]

EXCEPT! I think there's a bug in there where the everythingCollisionGroup defaults to 2147483648 (1000 0000 0000 0000  0000 0000 0000 0000) instead of -1 (1111 1111 1111 1111  1111 1111 1111 1111)

 

 

...or maybe that's smart thinking and lets you do bitwise "everything but" masks?