PDA

View Full Version : Hydraulic IK problems



Chrusion
04-01-2009, 02:57 PM
I am completely and utterly stupefied by what's happening in the attached super simple hydraulic test scene/object zip (scene is made in 9.3 so it can be loaded into all versions thereafter).

It goes like this... two (2) cylinder objects and one 360 deg bank rotating null. Layer 2 is the piston of a hydraulic cylinder and is parented to the null, offset as though attached to the rim of a wheel. IK is on bank and goal is simply Layer 1, the main cylinder. Conversely for the cylinder... simply has bank IK with its goal being the piston (yes, pivot point moved appropriate ends).

OK... with that, here's what happens...

In 9.3... diddly squat... no IK motion whatsoever, the piston just spins with the null and the cylinder is motionless.
In 9.5... the piston will point at its goal 10% of the rotation then spin with null the rest of the time, while the cylinder jutters around as it points towards it piston goal).
In 9.6... perfect fluid IK!

What the heck am I missing?

.

RebelHill
04-01-2009, 05:01 PM
ur missing the fact that the ik solver was changed from 9.3 to 9.5... and then further changes were made in 9.6... cross loading the same ik setup between the different versions will give differnt results

Chrusion
04-01-2009, 05:35 PM
Yeah, that's a given and understood. But the setup can't get any simpler than this! It's one object "looking" at another, created in 9.3, yet it doesn't work therein, but does in 9.6.

I mean, shouldn't any object rotate about it pivot when told to "go to a goal" (assuming proper IK params enabled and set)? What reason is there for an object to only partially rotate to point at its goal object or not rotate at all?

Again, what am I missing?

.

RebelHill
04-02-2009, 04:17 AM
item dependency issue most likely... Uve object 1 whos goal is object 2, and object 2 whos goal is object 1... in earlier version they cant "see" each other correctly. Ive seen this before in LW... for some reason, that i dont fully get, u can set something like this up, and it would work, but once u saved and reopened the scene it would all go wrong

Chrusion
04-02-2009, 09:36 AM
Sorry, I'm changing the point of this thread. Now the issue is not whether a 9.3 IK scene loads into 9.6, but the following...

The most basic and simplest of IK setups is this... one object and one null. Set the object's goal to be the null. Enable Full-time IK on the object. Set object's pitch, bank, OR heading controller to IK. Can't get any simpler.

It doesn't work in 9.3. The object DOES NOT rotate to point at the null.

Why?

.

RebelHill
04-02-2009, 11:49 AM
One object will require a child item... that child goes for the goal... the parent is rotated accordingly...

If ur trying to use IK as though it was "target", on just a single item with no children... then it wont work, and never has.

Chrusion
04-02-2009, 12:20 PM
ah HA! That's the key, Rebel! THANKS A TON!

Added a null to the end opposite each object's pivot, parented it to the respective object, set goal to respective object (default keyframe mode for controllers and enabled full-time IK), and finally set controller on objects to IK. Bingo! Happy hydraulic action.


Cylinder (pitch control = IK)
|
+-- CylinderEnd(null) -> Goal = Piston

Piston (pitch control = IK)
|
+-- PistonEnd(null) -> Goal = Cylinder



.

RebelHill
04-02-2009, 02:32 PM
ah HA! That's the key, Rebel! THANKS A TON!

pás problem, mon ami.

zapper1998
02-13-2012, 07:31 PM
cool