Rigging a board rolling over a ball (moving rotation pivots)?

[tom-hunt]

hi all

i'm trying to rig a board rolling over a ball, as seen in the screenshot. the board is parented to the control which encircles it. i would like the pivot of the board to always be in line with the ball as that is the center of rotation.

this is what i've been playing with:

boardCntrl.rotatePivotX = boardGEO.rotatePivotX = 0 - (boardCntrl.translateX-ballGRP.translateX);
boardCntrl.rotatePivotY = boardGEO.rotatePivotY = 0 - (boardCntrl.translateY-ballGRP.translateY);
boardCntrl.rotatePivotZ = boardGEO.rotatePivotZ = 0 - (boardCntrl.translateZ-ballGRP.translateZ);

the pivot is the negative of the board's translation so stays in the same position. the ball moves slightly with the board so to cancel this the ball's translation is subtracted at the end.

i'm sure this is not the best way to do it. any ideas?

thanks

tom

[oliver-magno]

This is very complex, the board's pivot should be moving along the surface while it is rotating, the point of contact of the board to the ball is not stationary, it is dynamically computed... the thing is... we need to find in which point is being in contact, then position the pivot of the board, then it should always tangent to the normal of the point of contact... whew!!! you gave me something to think... i'll do some experiments.

[joojaa]

The strange translation is beceause it double translates.

This is very complex

Not only that  you need to figure out a hierarchy that does not violate the dag and inst cyclic so thet there will be no too early bailouts by maya computation engine.

otherwise the math is simple its just basically a closet point on surface problem, which is well known mathematics, hell maya has a node that encapsulates this. However the
dag limitation is WAY more complex a problem.
So heres some toughts on this:

For starters the ball and plane must be siblings. Now then the trasform must be decoupled beacsue otherwise you end up in solve loop.

Or possibly you could do this by violating mayas computation mechanism and setting a setAttr in the expression, the downside of this is that you may see mayas animation capabilities disintegrationg.

Or if you can ensure a lock on the z move you could forego thsi problem also.

[joojaa]

oh blimey theres a better way-

What you need to do is have the trasfrom in a parent of the rotate and the board under that rotates transform. Then you need to matrix move the shape s that it trasfroms as if it was the board (because children can not act on their parents), Thsiway the translate is decoupled. Then its only basically a ik solve where the forst kinamatic chain ahs a fixed length and the rotation at elbow must be 90 degrees.

Must test at some point.