I am working in a new IP and is evaluating if we could use PhysX to replace Havok. Currently the trigger shapes do not work well with CCD, and bring us a lot of difficulties, when we are attaching trigger shapes on fast moving bullets. That’s really an important feature for us (and maybe for others, too). Do you think if that’s something you might consider to implement in the near future?

Thank you very much.

I would assume one would want it the other way around, where trigger shape would be attached to the volume the bullets are likely to enter. These would be static or slower moving, vs, have the volume enter the trigger shape of the bullet.

Please correct me if I was wrong. In my understanding the CCD would work (or not work) in both directions. Assuming the trigger shape is moving slowly, would the trigger event be generated if a fast-moving shape penetrating that? And after all, it’s difficult to guarantee at least one of the shapes are moving so slow that the trigger event could not be missed …

CCD doesn’t currently support “triggers” in the sense that the trigger shapes in PhysX are not processed by the CCD. However, you can emulate trigger-like behavior by making treating your triggers as regular simulation shapes and modifying your filter shader such that contact pairs involving your trigger pairs disable collision response and request touch found/lost events. This will give you a touch found event when the CCD or discrete simulation finds that the pairs come into contact and a touch lost event when the pair is lost.

Hope this helps