Constraint rest pose drifting

[AnonN10]

Let's say I have set up 2 bodies, lower and upper body (spheres). For a lower body I fix the horizontal axes of rotation, then for the upper body I bind it via slider or 6dof constraint to the lower body. Then I proceed to move the lower body around with AddForce. Why does the upper body slowly drift off its bind axis? I.e. doesn't come back to the position right above the lower body and instead to a position off-center relative to it.
Expected rest position:

Rest position after moving the lower body around for some time:

Settings for upper body

For lower body

Motor state is set to JPH::EMotorState::Position

[jrouwe]

Without a repro in the Samples application or at least a video it is hard to exactly tell you what is going wrong.

Some observations:

• Motor frequency is set to 0 which is not a supported value (should be > 0 and < 0.5 * simulation frequency) although you also have 'motor stiffness' and I don't know which ESpringMode your motor is in. It could be that the motor is not strong enough.
• The world pivot seems to be at the origin, I'm not sure where your objects are located but if the pivot is off it makes it harder for the constraint to work. For 2 equally heavy objects you'd want the pivot point to be in the middle for the best results.
• I don't know where 'limit min' and 'limit max' go but maybe not all of your 6DOF rotation axis are constrained and you don't say which motors you have activated so it could also mean that what you observe as translation is actually rotation. Maybe you should draw the constraint through Constraint::DrawConstraint to see what is actually going on.
• I've observed that restricting the EAllowedDOFs in combination with constraints can make the constraints less effective. You could try using another constraint to fixate the base instead.

[AnonN10]

I'll implement the debug rendering and post an update showing what it looks like in a video.

[AnonN10]

@jrouwe Just implemented that and noticed anchor points are at the origin, I wonder why...
PhysicsConstraintsDebugRenderer_crf35.webm
Could that be the cause of misalignment?

[AnonN10]

The solution ended up being configuring constraint in local space (EConstraintSpace::LocalToBodyCOM).