Hello again!
Me with my similar line of problems! Ok so to summarize again: I have a fbx hand model with bones and everything. Ive managed to set up the "wrists" and other joints with callback classes. These NodeCallback classes have access to an api that gets sensor data from the leap motion. We have a processing application that exposes positional data in an easy way to acccess and interpret. Each time the NodeCallback class is called I get the frame data: (pseudocode) Frame frame = OurAPI.frame()->getframe(); Now we have positional data. Ok now I want to move the fingers of my model to mirror my own fingers over top of a LEAP MOTION sensor. So provided the node we have is a bone: Code: osg::ref_ptr(osgAnimation::Bone) bone = dynamic_cast(osgAnimation::Bone*)(node); (side note, <> hides code in code brackets???) I want to rotate it on an axis relative to the position and rotation of the hand. The first thing I do is create a QUAT with the frame data using a vector 3 from each "joint" in teh hand. pseudocode: current = frame.finger.joints[currentjoint]; parent = frame.finger.joint[parentjoint]; child = frame.finger.joint[child]; Quat quat; quat.makerotat(current - parent, child - current); This way we calculate the rotation needed for angle at which the join should be rotated in relation to its parent and child joint. Correct? Then I need some Matrices. Matrixd ComputedRotation = ComputedRotation(quat) Matrixd rotationOfParent = makeRotate(parent bone rotation); Matrixd translationOfParent = makeTranslate (parent bone translation); Then I set the bone in skeletonspace bone->setMatrixInSkeletonSpace(ComputedRotation x RotationOfParent x translationOfParent x parent in skeletonspace); This way we move the matrix to the parent joint, then translate it back to where it should be relative to its parent, then rotate it to the same rotation of the parent and finally rotate it to how the frame data says it should. That gives quite a wild effect. The finger is bowed backwards and off to the side. I think it might be backwards because the axis are not correctly setup to match the osg axis but it shouldn't be so skewed. So if I remove the computedRotation then I would expect the finger to be placed back in its starting position (Aka how the hand model looked before I started moving it). But the finger is slightly off. I can't figure out for the life of me why. If I understand it correctly, the approach I am trying to take is forward kinematics. I have a rough grasp of it, but am getting very weird behaviours. Why is my finger always offset? How do I just get a joint to retain its relative position and affect only its rotation with a given angle? Thanks! ------------------ Read this topic online here: http://forum.openscenegraph.org/viewtopic.php?p=61117#61117 _______________________________________________ osg-users mailing list [email protected] http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
