Fast Automatic Skinning Transformations

How to read this

Category

Method: automatic skinning-transformation solver from sparse handles

Contributions

• Determines 2D and 3D skinning transformations from sparse user controls by minimizing as-rigid-as-possible elastic energies
• Expresses the energy purely in handle transformations and clusters vertices in weight space, running orders of magnitude faster than prior methods
• Enables new control modes such as shape-aware inverse kinematics and disconnected skeletons

Context

Builds directly on Jacobson et al.'s Bounded Biharmonic Weights, solving for the handle transformations over that weight space instead of supplying them manually.Builds on: Bounded Biharmonic Weights for Real-Time Deformation

Correctness

Speed comes from reducing the problem to handle transformations and clustering vertices into representative rotation groups, an approximation; quality therefore depends on the clustering and on the bounded-biharmonic weight basis it assumes, and the reported real-time rates are for the demonstrated meshes.

Clarity

Clearly written with concrete results; a first pass conveys the reformulation and its payoff, a second pass is needed for the rotation-clustering and optimization details.

How to read it

First pass for the reduce-to-handles idea and the new control modes; second pass on the weight-space clustering and ARAP minimization if you want the performance, since that is where the speedup lives.