-
A Rotation-Strain Method to Model Surfaces using Plasticity
Authors:
Jiahao Wen,
Bohan Wang,
Jernej Barbič
Abstract:
Modeling arbitrarily large deformations of surfaces smoothly embedded in three-dimensional space is challenging. The difficulties come from two aspects: the existing geometry processing or forward simulation methods penalize the difference between the current status and the rest configuration to maintain the initial shape, which will lead to sharp spikes or wiggles for large deformations; the co-d…
▽ More
Modeling arbitrarily large deformations of surfaces smoothly embedded in three-dimensional space is challenging. The difficulties come from two aspects: the existing geometry processing or forward simulation methods penalize the difference between the current status and the rest configuration to maintain the initial shape, which will lead to sharp spikes or wiggles for large deformations; the co-dimensional nature of the problem makes it more complicated because the deformed surface has to locally satisfy compatibility conditions on fundamental forms to guarantee a feasible solution exists. To address these two challenges, we propose a rotation-strain method to modify the fundamental forms in a compatible way, and model the large deformation of surface meshes smoothly using plasticity. The user prescribes the positions of a few vertices, and our method finds a smooth strain and rotation field under which the surface meets the target positions. We demonstrate several examples whereby triangle meshes are smoothly deformed to large strains while meeting user constraints.
△ Less
Submitted 8 August, 2022;
originally announced August 2022.
-
A Deep Emulator for Secondary Motion of 3D Characters
Authors:
Mianlun Zheng,
Yi Zhou,
Duygu Ceylan,
Jernej Barbič
Abstract:
Fast and light-weight methods for animating 3D characters are desirable in various applications such as computer games. We present a learning-based approach to enhance skinning-based animations of 3D characters with vivid secondary motion effects. We design a neural network that encodes each local patch of a character simulation mesh where the edges implicitly encode the internal forces between th…
▽ More
Fast and light-weight methods for animating 3D characters are desirable in various applications such as computer games. We present a learning-based approach to enhance skinning-based animations of 3D characters with vivid secondary motion effects. We design a neural network that encodes each local patch of a character simulation mesh where the edges implicitly encode the internal forces between the neighboring vertices. The network emulates the ordinary differential equations of the character dynamics, predicting new vertex positions from the current accelerations, velocities and positions. Being a local method, our network is independent of the mesh topology and generalizes to arbitrarily shaped 3D character meshes at test time. We further represent per-vertex constraints and material properties such as stiffness, enabling us to easily adjust the dynamics in different parts of the mesh. We evaluate our method on various character meshes and complex motion sequences. Our method can be over 30 times more efficient than ground-truth physically based simulation, and outperforms alternative solutions that provide fast approximations.
△ Less
Submitted 11 April, 2021; v1 submitted 1 March, 2021;
originally announced March 2021.
-
Modeling of Personalized Anatomy using Plastic Strains
Authors:
Bohan Wang,
George Matcuk,
Jernej Barbic
Abstract:
We give a method for modeling solid objects undergoing large spatially varying and/or anisotropic strains, and use it to reconstruct human anatomy from medical images. Our novel shape deformation method uses plastic strains and the Finite Element Method to successfully model shapes undergoing large and/or anisotropic strains, specified by sparse point constraints on the boundary of the object. We…
▽ More
We give a method for modeling solid objects undergoing large spatially varying and/or anisotropic strains, and use it to reconstruct human anatomy from medical images. Our novel shape deformation method uses plastic strains and the Finite Element Method to successfully model shapes undergoing large and/or anisotropic strains, specified by sparse point constraints on the boundary of the object. We extensively compare our method to standard second-order shape deformation methods, variational methods and surface-based methods and demonstrate that our method avoids the spikiness, wiggliness and other artefacts of previous methods. We demonstrate how to perform such shape deformation both for attached and un-attached ("free flying") objects, using a novel method to solve linear systems with singular matrices with a known nullspace. While our method is applicable to general large-strain shape deformation modeling, we use it to create personalized 3D triangle and volumetric meshes of human organs, based on MRI or CT scans. Given a medically accurate anatomy template of a generic individual, we optimize the geometry of the organ to match the MRI or CT scan of a specific individual. Our examples include human hand muscles, a liver, a hip bone, and a gluteus medius muscle ("hip abductor").
△ Less
Submitted 2 August, 2020;
originally announced August 2020.