← ArchivePaper2023

3D Mesh Pose Transfer Based on Skeletal Deformation

Shigeng Yang, Mengxiao Yin, Ming Li, Guiqing Li, Kan Chang, Feng Yang

Edge-convolution network extracts skeleton and rigid weights, then diffuses smooth weight fields before deforming source mesh to a reference pose.

Open the paper ↗Trace in the lineage graph

Abstract

For 3D mesh pose transfer, the target model is obtained by transferring the pose of the reference mesh to the source mesh, where the shape and pose of the source are usually different from that of the reference. In this paper, pose transfer is considered as a deformation process of the source mesh, and we propose a 3D mesh pose transfer method based on skeletal deformation. First, we design a neural network based on the edge convolution operator to extract the skeleton of the 3D mesh and bind the rigid weights; then, we calculate the bone transformations between the two skeletons with different poses and use the diffusion equation to smooth the rigid weights; finally, the source mesh is deformed according to the bone transformations and the smooth weights to get the target mesh. Experiment results on different datasets show that the pose of the reference mesh can be effectively transferred to the source one while maintaining the shape and high‐quality geometric details of the source mesh by using our method.

How to read this

Category: Method: a skeleton-based mesh pose-transfer algorithm
Contributions: An edge-convolution network that extracts a 3D mesh skeleton and binds rigid skinning weights
A diffusion-equation step that smooths the rigid weights before deformation
Pose transfer cast as a deformation driven by computed bone transformations plus smoothed weights, preserving source shape and detail
Context: Relates to classic deformation-driven transfer such as Sumner's Deformation Transfer for Triangle Meshes, recasting the problem through learned skeleton extraction and skinning rather than direct correspondence.Builds on: Deformation Transfer for Triangle Meshes
Correctness: Demonstrated on different datasets showing pose is transferred while source shape and geometric detail are kept; the skeletal/rigid-weight assumption may limit fidelity for highly non-rigid or topology-mismatched cases, which a reader should keep in mind.
Clarity: Reasonably accessible; a first pass gives the three-stage pipeline, a second pass is needed for the edge-convolution and diffusion details.
How to read it: Trace the pipeline stage by stage (skeleton extraction, weight smoothing, deformation); a second pass on the diffusion smoothing pays off if weight quality matters to you.

Builds on

Deformation Transfer for Triangle Meshes 2004

Built upon by

Nothing yet.

Related work

Delta Mush: Smoothing Deformations While Preserving Detail 2014 / DigiPro
Implicit Skinning: Real-Time Skin Deformation with Contact Modeling 2013 / SIGGRAPH
Skinning: Real-time Shape Deformation 2014 / SIGGRAPH
Expressive Body Capture: 3D Hands, Face, and Body from a Single Image 2019 / CVPR

Keywords

pose transfer skeletal deformation linear blend skinning skeleton extraction rigid skinning weights edge convolution gkedgeconv dgcnn joints regression diffusion equation cotangent laplacian bone transformation smpl smal neural blend shape mesh deformation detail-preserving

This page summarises the entry and links to its original source. The archive never hosts or redistributes the publication itself.Show it in the full archive list →