Physics Asset Constraint not working as expected

I made a video to show and explain the issue I’m running into

But basically the constraint in my physics asset isn’t behaving how the indicator or settings suggest it should. It appears that the simulation is ignoring the constraint orientation altogether, because although the limits are applied, they are applied with no regard to the constraint orientation, as if they are using the default bone orientation or something no matter what.

This is in 4.25.4. The constraints work properly on a sample ragdoll project that uses the UE mannequin model/skeleton, so it’s not universally broken, but there is certainly something about our skeleton that it isn’t handling properly. I can’t find any documentation or otherwise that place any restrictions on how the skeletal bones should be oriented though, and there is no editor indicators that point me towards the potential problem, so I’m at a loss. I’ve filed a bug that includes one our our models, through the editor issue reporting tool, but that tends to be a 1 way black hole.

Someone on the Unreal Slackers Discord group pointed me in the right direction. Apparently when you are rotating the constraint with the rotation tool, even though the widget arcs are rotating, leading you to expect that the constraint should operate within that visual arc, that’s not what is actually happening. You need to use ALT+Rotate tool in order to rotate the “frame of reference” without also rotating the constraint. I don’t fully understand the relationship between the two at the moment, but as a user I feel like there are still issues with this work flow. The normal rotate tool shouldn’t rotate the arcs if they are not ultimately effecting the arcs. When the constraint limits visuals(arcs) changes from a normal rotation operation, but the constraint doesn’t reflect that visual, it still comes across as a bug, if only in the rendering of the widgets. If a regular rotation of the constraint isn’t actually effecting the limits, the limits shouldn’t move with that rotation, but only with an ALT+Rotate operation.

Some clarification of the widget rendering would be really nice here, as would some actually useful documentation. The page on editing constraints in PHAT make no mention of an alternate rotation mode, or the presence of more than 1 rotation type for constraints…nts/index.html

TLDR: The editor and documentation does a terrible job of communicating the presence of 2 distinct rotational modes for constraints, and the editor constraint widget doesn’t reflect a “what you see is what you get”, making them appear broken. There is still a bug in here somewhere. This workflow is bad.


Happy to have helped you and thanks for summarizing the solution to the benefit of the community!

Thank you. I would have never worked this one out.

Wow, I wasted so much time experimenting with this. Thanks for that. This is not obvious at all.