Eva Schiffer

Footskate Cleanup for Motion Capture Editing

Summary:

This paper discusses a system for enforcing foot planing constraints on motion capture data in a less expensive way by allowing small changes to the size of the underlying skeleton. Unfortunately the usefulness of these warped skeleton motions is not evaluated from a viewer perception slant, so it is difficult to know if this system would create visibly smooth motions.

Problem:

The authors wanted to make it easier to correct the problem of sliding feet in motion capture data by providing a tool that would be able to enforce numerous foot plant constraints, but still remain efficient and only deform the skeleton in the most minimal ways. 

Method:

The system described by the author uses a relatively simple skeleton system for the lower legs in order to keep the adjustment of the bones simple and knowledge about foot placement on the floor to help constrain the problem of where to place the rest of the body.
The system handles foot plant constraints by recording when and where foot plants occur, then adjusting the rotation and orientation of the bones in order to get as close to the correct foot plant as possible. Then the system makes up for the remaining distance by adjusting the bones in the leg to make up the extra small distance. The final displacement at a time stamp (when a foot is planted) is then dissipated over time so that the placement will blend with the rest of the motion.

Key Ideas:

The most important ideas in this paper are:

	1. Foot plants can be enforced.
	2. Small relaxations of fundamental constraints can make complex mathematical problems easier (or possible) to solve.
	3. Higher level control of motion capture data is possible, but still requires specially tuned tools.

Contributions:

This paper contributed a powerful footskate cleanup program as well as exploring how to do complex motion capture editing.

Questions:

I am curious how realistic the motions that resulted from this system are?
Would people normally see the stretching or would the extra joint bending (that would be needed without the stretching) be more notable?