- bunch of guidelines for how to control a camera...
- often want to specify what the camera 'looking at'; then animate it along a path and let the computer ensure it looks at its target throughout the animation
- (Armatures/Rigging in Blender)
- Give a 3d model bones/joints separate from the model (not hierarchical like puppet)
- Manipulate the (simpler) skeleton to manipulate the (complex) model
- Also useful for previewing animations without having to manipulate the complex model at every frame
- Skeleton not visible during render
- Embed the object in a bounded grid; transforming the control points of the grid causes the object to transform accordingly
- Ex: https://www.youtube.com/watch?v=Bhilhg0rQtY
- forward kinematics: Given skeletal transformations, transform model
- inverse kinematics: Given model transformation (ex: I want the hand to be at (0,0,0)), find skeletal transform to achieve it (what angles should all the joints be at to achieve the hand in the given position)
- Forward Dynamics: Given forces, determine model
- Inverse Dynamics: Given desired state, determine necessary forces
- only briefly covered the math
- I think this is similar to FFD, except the grid has 'springs' between each of the control points (also masses here), and you apply forces to the grid
- Walking is hard
- Walking realism follows 'uncanny valley' graph (not proven): See #file-uncanny-valley-svg
- links: like bones
- sensors:
- touch sensors: check if touching something (i.e. avoid foot going through the ground)
- angle sensors: joints
- actuators: sort of like constraints? On angles (joints) or length (stretchiness?)
- Simulate how muscles contract; minimize metabolic energy to improve walking
- Ex: https://www.youtube.com/watch?v=r_c66EZAqrI
