Techniques for Motion Editing is a detailed analysis of the concepts and methods behind motion editing, as well as a look into some of the difficulties involved with editing motion capture data.  Motion editing is necessary to clean up or adjust motion capture information that may contain imperfections or does not necessarily give the animator the movement they require for a particular scene or circumstance.  Also, motion capture is not always capable of re-producing the exact emotional effects desired by the animator, possibly due to the limitations of the technology.  Motion editing allows for the changing of virtually any part of the captured motion.

Key ideas:

1. Poses can be represented as either functions of position at a given time, or represented by how specific instants of position change over time; the preferred method for motion capture character animation is a skeleton of rigid pieces (bones) that can be parameterized by their position and orientation.  

2. Rotations of a rigid body can be represented in several different ways- Euler angles, unit Quaternions, rotation matrices, exponential maps- each with their own set of benefits and shortcomings.  

3. Motion capture is a sampled representation of motion since it only tells us what the parameters of a rigid body are at discrete time intervals.  Parameter curves were introduced as a useful method of representing position over time, however motion editing is made somewhat more complex by this dense representation of data.

4. When editing motion to clean up unwanted noise, one must be mindful of the effect a low-pass filter has of having a tendency to destroy "crisp" edges to a movement that the animator may wish to preserve.  

5. Key reduction is the process of converting raw motion capture data into a smaller set of representative frames.  This method is incredibly beneficial for motion editing since it is much harder to control and modify the motion capture data compared to reduced number of key frames that still preserve the original data.

6. Parameter curve representation allows us to characterize motion as a signal, which in turn gives us motion analysis tools and a new set of operations that we can apply to our motions.