Avatar Reshaping and Automatic Rigging Using a Deformable Model

How to read this

Category

System: automatic rigging and reshaping of body scans into animatable avatars

Contributions

• Fitting a simplified SCAPE-based morphable model with linear blend skinning to a single 3D body scan via ICP-style pose and shape optimization
• Skeleton joint transfer using mean-value coordinates and skin weight propagation by harmonic interpolation
• Interactive reshaping along semantic attributes such as height and weight via body-shape-space exploration and deformation transfer

Context

Combines morphable-model fitting with auto-rigging in the lineage of 'Automatic Rigging and Animation of 3D Characters' (Pinocchio), using a SCAPE-based deformable model as the prior.Builds on: Automatic Rigging and Animation of 3D Characters

Correctness

Quality of rig and reshaping rests on how well the SCAPE-based template fits the input scan; the work reports superior skin deformation versus generic single-mesh auto-rigging like Pinocchio, but it targets human bodies from clean single scans, so results outside that regime are not claimed.

Clarity

Accessible as a system description; a first pass conveys the pipeline, a second pass details each fitting and transfer step.

How to read it

Read as a pipeline: first pass for the fit then transfer then reshape flow; do a second pass on the joint and weight transfer if you want to reproduce the rigging quality.