inherit rotation problem (bug?)

[Anonymous]

Hi all – I’ve a hierarchy problem which I think is a bug – suggestions (polite ones) welcomed…

I want one object to rotate related to another, but only about one axis. To do this I’ve used two dummies – one linked to the other. Then on the child object/dummy I’ve removed all of the hierarchical inherit links, except for its z rotation. This appears to work fine when the dummies are exactly placed on top of eachother.

But when I position the child object away from the parent it starts to show inherit rotations when the parent is rotated about its x or y axis – which is exactly what I do Not want.

I’ve included a two simple max files to illustrate this. Ie. Rotate the parent dummy on each axis and watch the child dummy jump around. Does anyone know how I can fix this?

[Brock_Lafond]

Away from workstation right now and thinking out loud.

Could you not achieve this using wire parameters? Wire the axial rotation of one dummy to the other.

[Anonymous]

Thanks for the quick reply - what I actually need to make happen is for one object to rotate about one axis relatvie to another which moves around. For this I was going to use an object with a look at controller - and child linked to the object with its axis rotation constrained - I've tried a range of setups and none see to work.

picture boxes rotating flat on the ground so they face another object which flies around...

The example I've given are for a inherit link setup which should imo work, but doesn't seem to.

[Brock_Lafond]

Thinking out loud again.

Create a bone that aligns with the dummy that you want to rotate and create an end nub. Add a history independent IK solver from the bone to the nub. In the IK solver properties set the swivel angle to use the other dummy as the target. Wire the swivel angle amount to the axis of the dummy you want to rotate.

Away from my desk so I can't test this, but it's fun to think about.

Edit: If the rotation seems to be too strong, try using DegToRad( _ rotation)

[Steve_Curley]

No, it's not a bug - you're missing the point with inherited rotation. Let's say you have DummyB linked to DummyA - when DummyA is rotated, in any direction, DummyB will follow that rotation as if its Pivot Point was aligned to the Pivot of DummyA. So yes, if it lies directly along one of the main axes with respect to DummyA it will appear to be rotating about it's own pivot, but it isn't. That's why it "creeps" the further away from that axis (on DummyA) it is.

Try this:-
Create Dummy A and DummyB as before, apply a LookAt Constraint to DummyB using DummyA as the Target. Create another (slightly larger) one - DummyC - Align it to DummyB and Select and Link it to DummyB. Deselect X and Y in the Inherit Rotation section of the Link Info. Voila! A Dummy which always faces the moving object but always stays flat on the ground.

[Anonymous]

Thanks for the replies.

My original problem came from a setup which was more complicated than the examples I gave. This had several linked and attached objects which moved the childs pivot away from its parent. I’ve now reset pivots so they align and the problem is corrected.

Although I’d like to suggest that the example I gave still shows an error – if you take the ‘test2-offset problem’ I’ve attached and rotate the parent dummy about its local y or z axis you’ll notice the child begins to flip suddenly when the parents rotation passes through 90 degree values (would this not be called a bug?).

The parent is using Bezier float for its rotations, so I’d suggest the problem relates to that – it would be nice to have it working as expected though, through any rotation/position.

However, with pivot points aligned I can continue.

Thanks again for your feedback.

Dave