Transforming SOPs before Actor COMP causing weirdness?

Hey all!
Diving into bullet solver stuff recently, and having a lot of fun with it.

I’m running into what seems like odd behavior, so wanted to see if anyone else can confirm this is incorrect or expected behavior:

I have a box sop in my actor comp, followed by a transform sop, pushing it down on Y by 1 unit. Then I have a null sop with render flags on which the parent actor comp picks up as the collision geometry:

Then I make a constraint comp, and without any modifications really , out of the box it has the pivot of the constraint raised by 1 unit vertically, opposite direction of the sop transformation.

If I displace the cube up instead of down in the actor comp, the constraint pivot goes the other way, the inverse.

Now, if I cancel out the transform sop totally and bypass it, then transform the actor comp up a unit, it looks as I’d expect:

Anyone have any ideas if this is a bug, or intended behavior?

Thanks!

Sounds to me like it may be a bug, either in the constraint viewer or the constraint itself. Any idea if what you see in the constraint viewer accurately reflects the constraint?

If you’re modifying the underlying SOPs it could also be that the collision shapes are out of date and need to be updated.

Hey @eric.b ,

collision shapes are updated for sure.

I think it has to do with the constraint itself, not the viewer representation, though I’m not so sure as the actor reacts strangely, it also has some weird axis skewing that causes the actor to rotate on a plane that changes or is really odd. hard to explain, so attaching a similar file here:

constraintWeirdness.1.toe (4.5 KB)

That is quite strange indeed. Thanks for the example, I’ll take a look into it.

At the very least there is an issue with the viewer because the hinge should not be moving with the collision shape.

1 Like

That makes sense! Thanks for taking a look into this.

I identified and fixed two issues:

  • the viewer guides were not matching the constraint and were moving when they shouldn’t be
  • there were cases where the hinge constraint was not on the correct axis of rotation, like in your attached toe

These fixes will be in the next official build we release. Thanks for the report.

woo! awesome excited to dig into that. Thanks again :slight_smile: