1. This paper presents an automated framework that allows character artists to use the full complement of tools in high-end systems to create characters for interactive systems.

2. Convincing skin deformations are typically created in commercial animation packages which provide very general and powerful tools. While these systems are convenient and fexible for artists, the generality often lea$ds to characters that are slow to compute or that require a substantial amount of memory and thus cannot be used in interactive systems. Instead,interactive systems restrict artists to a specific character deformation model which is fast and memory efficient but is notoriously difficult to author and can suffer from many deformation artifacts.

3. This method starts with an arbitrarily rigged character in an animation system. A set of examples is exported, consisting of skeleton configurations paired with the deformed geometry as static meshes. Using these examples, we fit the parameters of a deformation model that best approximates the original data yet remains fast to compute and compact in memory.

4. Key techniques
4.1 Extend the linear blending algorithm
The traditional linear blend skinning model is extended by adding a relatively small number of joints that are simply related to the original skeletal parameters and fit using them. We choose these extra joints by both examining the places where the standard linear blend model fails and by examining extra character deformations that we would like to capture. We then add joints that we believe will help resolve these artifacts. Finally, we fit the parameters of our skinning model using this extended skeleton.

4.2 Fit the skinning model
Step 1: Finding Influence Sets
Measure how rigidly a vertex transforms with every joint over all examples and use the most rigidly transforming joints for the influence set.

Step 2: Solving for Weights and Vertices
Find the best vertices and weights that minimize the least-squares difference between the skin and the examples at all the example skeleton configurations.

Step 3: Handling Normals

5.Limitation and Future improvement
5.1 Character deformations are only driven by the skeleton's joint parameters,and can not capture deformations that are driven by abstract parameters such as "happiness".

5.2 Cannot accurately reproduce deformations that are not representable as linear combinations of the transformations expressed in skeletons.