How to rotate constraints

[dxhhx]

Hopefully this clip explains the problem:

2 packed geos are hard constrained together, tube animation (translate and rotate) is transferred from SOPs. The constraint works for position but rotation, how to make sure the constraint network understands there is rotation?

Desired behaviour:

Scene file:

001.hip

 

 

desired.webm

current.webm

Edited October 24, 2021 by dxhhx

[Librarian]

@dxhhx I have this.. maybe it will be useful.

 

animROT-constraints.hiplc

[dxhhx]

@Librarian Thanks for the file! That was indeed helpful

I'm still curious to see if there is a way for the constraints to infer their orientation from the rotation of the geometry it's anchored to, like in the file I uploaded

 

 

[julian johnson]

001a just turns off your Geometry wrangle and switches the object to animated static which allows the Dopnet to update the geometry orientation and in turn the constraint.

001c adjusts only your Geometry wrangle to also update the orient attribute so that the constraint can 'see' the rotation.

Not sure on the ins and outs of why these should 'work' but they seem to do what you want. Maybe?
J

001a.hipnc

001c.hipnc

[dxhhx]

@julian johnson this is exactly what I was looking for. Thank you!!

[dxhhx]

So one follow-up question: the constraint drifts in time, I've made the constraint longer to make it more obvious (desired behavior the link points and anchors are collinear):

Increasing the "Error Reduction Parameter" inside the Hard Constraint Relationship node makes it better but it's not perfect

Is there some other setting that would fix this?

002.hip

 

 

Edited October 24, 2021 by dxhhx

[toadstorm]

I'm honestly not entirely sure why the hard constraint isn't enforcing rotation well enough, but the cheap workaround would be to use a Glue Constraint Relationship instead. Set s@constraint_type to "glue" with a f@strength of -1, then use a Glue Constraint Relationship DOP in place of your Hard Constraint. This will force both objects to solve as a single physical object.

You also don't need multiple DOP objects for this system to work. As long as you set the i@active and i@animated attributes correctly on your static and active pieces, you can solve everything as a single DOP object and simplify things a little.

002_toadstorm.hip

[dxhhx]

Hey Henry!

Just saw this. I was trying stuff earlier and found if I use 2 constraints, one "position" type and another "rotation", it works a lot better than using just one set to "all". Very bizarre:

 

 

I'm going to check your file out now

[dxhhx]

3 hours ago, toadstorm said:

I'm honestly not entirely sure why the hard constraint isn't enforcing rotation well enough, but the cheap workaround would be to use a Glue Constraint Relationship instead. Set s@constraint_type to "glue" with a f@strength of -1, then use a Glue Constraint Relationship DOP in place of your Hard Constraint. This will force both objects to solve as a single physical object.

You also don't need multiple DOP objects for this system to work. As long as you set the i@active and i@animated attributes correctly on your static and active pieces, you can solve everything as a single DOP object and simplify things a little.

002_toadstorm.hip

Yep glue works great. interesting.

Thanks Henry

[toadstorm]

1 hour ago, dxhhx said:

Hey Henry!

Just saw this. I was trying stuff earlier and found if I use 2 constraints, one "position" type and another "rotation", it works a lot better than using just one set to "all". Very bizarre:

 

 

I'm going to check your file out now

There must be a good reason why this works, but hell if I know what it is. Nice find!

Edited October 25, 2021 by toadstorm