Skinning on Progressive Decimated Models

Abstract

Skinning existing animation frames or examples is exploited actively to find the most suitable scheme that is capable of capturing deformations from given mesh sequences. It is necessary to approximate joint transformations by a fitting algorithm that usually involves solving a large scale linear system. In this paper, we reduce the dimensions of the linear system substantially by representing example meshes as progressive decimated models. These models are reconstructed from augmented deformation sensitive decimation (ADSD) that is able to handle large deformations while maintaining connectivity relationship throughout all example meshes. We also show that skinning can be propagated to decimated models, which allows animations to be scalable to specific applications with varied requirements of qualities.

Skeletal character animation from examples

“…The wire curves give the definition of an object shape and outline its deformable features. This technique mimics the usage of armatures for sculptors and is widely practiced in industries [2]. Lazarus et al propose axial deformations, in which a 3D axis (represented as a 3D curve) is built and positioned by users for a shape.…”

“…We need to expand Rpk to homogeneous matrices and then convert them into quaternions, which are trivial works. However blending quaternions are somewhat complicated even though a straightforward linear blending is performed in [2]. The algorithm is described as following:…”

“…There are many works dedicated to decimating a single mesh, but very few of them consider a group of example meshes or take into account of varying deformations presented in different examples. The target of this chapter is thus to provide a representation scheme for the proposed methods in the previous chapters.The classic decimation scheme used for a group of examples is improved to deal with character examples that usually contain very large deformations.The work in this chapter is published in[149,151]. The implementation is performed on a PC with Dual Core 2.66 GHz, 2G RAM and a Quadro FX 560 Graphic Card.Statistics of approximating joint transformations for a progressive horse model at one single specific pose with varied resolutions are given inTable 7.1.…”

Direct rendering of feature-based skinning deformations

“…Moreover, their uniform mesh clustering often leads to unpleasant artifacts on articulated Prior Art Clustering Method Feature Application [131] Region Growing Euclidean & Angular distances Metamorphosis [124] Clustered PCA ⋆ Euclidean distance Compression [155] Clustered PCA ⋆ Euclidean distance Editing [32] Region animations and highly detailed animated regions because it does not attempt to capture important parts of the animation. Subsequently, Xian et al [163] reduced the dimensions of the SAD linear system substantially by representing animated meshes as progressive decimated models. As a result, the fitting time is reduced at the expense of a significant increase of the fitting error.…”

“…Similarly to segmentation, key-frames extraction methods of animation sequences can also be used 132 to reduce the dimensions of the fitting optimization. Mesh simplification algorithms [163] can also be used for the same reason. Finally, we believe that temporal coherence for deforming objects deserves further investigation.…”