phy.add contact callback is called for all contacts when using Havok

[fatbatman]

phy.add({ type:'contact', b1:obj1.name, b2:'obj2.name, callback: someCallback })

In Havok someCallback is called for all collisions involving obj1, not just those involving obj2 but arg.hit is always false.

Phsyx works as expected

Rapier, seems to always have arg.hit == true

Also, including all details of the contact in the collision callback argument is probably needed in most real world uses - body1, body2, position, velocity etc

[fatbatman]

Are there any work arounds for this?

[lo-th]

yep i will make demo about that
and standardize the process for all engine

[brandon-lb]

Actually, for me Havok worker "Character" example never calls any trigger. You can verify this by adding console logs to showContact and triggerContact and switching between PhysX and Havok.

[brandon-lb]

By the way, I actually want to get a list of all items that are colliding with a given mesh. Any idea the best way to do that?

[brandon-lb]

Sorry to pile on, but I also noticed both the point and normal arrays on a hit are always [0,0,0] as well.

[fatbatman]

What's the recommended way to access the underlying physics engine so I can register a contact listeners directly as a workaround for now?

[lo-th]

contact is add to havok
trigger is not constant ? i search why
collision return need more research but is in good way

[fatbatman]

It work better In the latest version with Havok - arg.hit is set correctly

I noticed b1, b2 must be in a specific order though

phy.add({ type:'contact', b1:'box1', b2:'sensor1', callback: trigContact}) //is called and arg.hit is set correctly
phy.add({ type:'contact', b1:'box2', b2:'sensor1', callback: trigContact}) //is called and arg.hit is set correctly

//This behaves differently:
phy.add({ type:'contact', b1:'sensor1', b2:'box1', callback: trigContact}) //is called and arg.hit is set correctly
phy.add({ type:'contact', b1:'sensor1', b2:'box2', callback: trigContact}) //# IS NEVER CALLED #

contact point and normal information still isn't present.

Some way of register for all contacts of a given object using a wildcard would be useful.

[fatbatman]

Using Havok contact events don't get triggered after registering more than about ~146.
This in a scene with about 160 rigid bodies in total.

[lo-th]

mm ok i can increase contact

[fatbatman]

Is it possible to have wildcard for all contacts from an objects?
as currently I have to register an object, e.g. a bullet, against all possible targets.
Also an ability to remove the contact.

[lo-th]

problem is with worker i have to limite message and return array
i will think about that

[fatbatman]

I see you increased the number of contacts that can be supported.
This means it lasts longer, but then runs out as old ones never get cleaned up.
However, I can remove the contacts based on their name, but it means I need to store a mapping of those contact pairs relevant to each body.