Efficient Penetration Depth Approximation using Active Learning

Jia Pan, Xinyu Zhang, Dinesh Manocha

We present a new method for efficiently computing the global penetration depth between two rigid objects using machine learning techniques. Our approach consists of two phases: offline learning and performing run-time queries. In the learning phase, we pre-compute an approximation of the contact space of a pair of intersecting objects from a set of samples in the configuration space. We use active and incremental learning algorithms to accelerate the pre-computation and improve the accuracy. During the run-time phase, our algorithm performs a nearest-neighbor query based on translational or rotational distance metrics. The run-time query has a small overhead and computes an approximation to global penetration depth in a few milliseconds. We use our algorithm for collision response computations in Box2D and Bullet game physics engines and observe more than an order of magnitude improvement over prior PD computation techniques.

Efficient Penetration Depth Approximation using Active Learning

A Material Point Method for Snow Simulation

Alexey Stomakhin, Craig Schroeder, Lawrence Chai, Joseph Teran, Andrew Selle

Snow is a challenging natural phenomenon to visually simulate. While the graphics community has previously considered accumulation and rendering of snow, animation of snow dynamics has not been fully addressed. Additionally, existing techniques for solids and fluids have difficulty producing convincing snow results. Specifically, wet or dense snow that has both solid- and fluid-like properties is difficult to handle. Consequently, this paper presents a novel snow simulation method utilizing a usercontrollable elasto-plastic constitutive model integrated with a hybrid Eulerian/Lagrangian Material Point Method. The method is continuum based and its hybrid nature allows us to use a regular Cartesian grid to automate treatment of self-collision and fracture. It also naturally allows us to derive a grid-based semi-implicit integration scheme that has conditioning independent of the number of Lagrangian particles. We demonstrate the power of our method with a variety of snow phenomena including complex character interactions.

A Material Point Method for Snow Simulation

Course: Turbulent Fluids

Tobias Pfaff, Nils Thuerey, Theodore Kim

Over the last decade, the special effects industry has embraced physics simulations as a highly useful tool for creating realistic scenes ranging from a small camp fire to the large scale destruction of whole cities. While fluid simulations are now widely used in the industry, it remains inherently difficult to control large scale simulations, and there is an constant struggle for increasing visual detail.

In this course, we will tackle these problems using turbulence methods. Turbulent detail is what makes typical fluid simulations look impressive, and the underlying physics motivate a powerful approach for control: they allow for an elegant split of large scale motion and small scale turbulent detail. This results in a two-stage work flow that is highly convenient for artists: first, a rough, and fast initial simulation is performed, which is then turned into a more detailed one by adding turbulent effects.

This course aims at giving an overview and providing practical guide to employing turbulence modeling techniques for fluid simulations in computer graphics. After reviewing the basics of fluid solvers, and the popular wavelet turbulence approach, we will present several powerful methods to capture advanced effects such as boundary layers, and turbulence with directional preferences. In addition, the difficulties of liquid simulations will be explained, and an approach for liquid turbulence that is based on wave dynamics will be presented.

Turbulent Fluids

SIGGRAPH Asia 2013

Ke-Sen’s full list here. Without further ado:

A New Grid Structure for Domain Extension

Bo Zhu, Wenlong Lu, Matthew Cong, Byungmoon Kim, Ronald Fedkiw

We present an efficient grid structure that extends a uniform grid to create a significantly larger far-field grid by dynamically extending the cells surrounding a fine uniform grid while still maintaining fine resolution about the regions of interest. The far-field grid preserves almost every computational advantage of uniform grids including cache coherency, regular subdivisions for parallelization, simple data layout, the existence of efficient numerical discretizations and algorithms for solving partial differential equations, etc. This allows fluid simulations to cover large domains that are often infeasible to enclose with sufficient resolution using a uniform grid, while still effectively capturing fine scale details in regions of interest using dynamic adaptivity.

A New Grid Structure for Domain Extension

A Level Set Method for Ductile Fracture

Jan Hegemann, Chenfanfu Jiang, Craig Schroeder, Joseph M. Teran

We utilize the shape derivative of the classical Griffith’s energy in a level set method for the simulation of dynamic ductile fracture. The level set is defined in the undeformed configuration of the object, and its evolution is designed to represent a transition from undamaged to failed material. No re-meshing is needed since the resulting topological changes are handled naturally by the level set method. We provide a new mechanism for the generation of fragments of material during the progression of the level set in the Griffith’s energy minimization. Collisions between different material pieces are resolved with impulses derived from the material point method over a background Eulerian grid. This provides a stable means for colliding with embedded interfaces. Simulation of corotational elasticity is based on an implicit finite element discretization.

A Level Set Method for Ductile Fracture

Efficient Simulation of Secondary Motion in Rig-Space

Fabian Hahn, Bernhard Thomaszewski, Stelian Coros, Sebastian Martin, Robert Sumner, Markus Gross

We present an efficient method for augmenting keyframed character animations with physically-simulated secondary motion. Our method achieves a performance improvement of one to two orders of magnitude over previous work without compromising on quality. This performance is based on a linearized formulation of rig-space dynamics that uses only rig parameters as degrees of freedom, a physics-based volumetric skinning method that allows our method to predict the motion of internal vertices solely from deformations of the surface, as well as a deferred Jacobian update scheme that drastically reduces the number of required rig evaluations. We demonstrate the performance of our method by comparing it to previous work and showcase its potential on a production-quality character rig.

Efficient Simulation of Secondary Motion in Rig-Space

Subspace Integration with Local Deformations

David Harmon, Denis Zorin

Subspace techniques greatly reduce the cost of nonlinear simulation by approximating deformations with a small custom basis. In order to represent the deformations well (in terms of a global metric), the basis functions usually have global support, and cannot capture localized deformations. While reduced-space basis functions can be localized to some extent, capturing truly local deformations would still require a very large number of precomputed basis functions, significantly degrading both precomputation and online performance. We present an efficient approach to handling local deformations that cannot be predicted, most commonly arising from contact and collisions, by augmenting the subspace basis with custom functions derived from analytic solutions to static loading problems. We also present a new cubature scheme designed to facilitate fast computation of the necessary runtime quantities while undergoing a changing basis. Our examples yield a two order of magnitude speedup over full-coordinate simulations, striking a desirable balance between runtime speeds and expressive ability.

Subspace Integration with Local Deformations

Thin Skin Elastodynamics

Duo Li, Shinjiro Sueda, Debanga R. Neog, Dinesh K. Pai

We present a novel approach to simulating thin hyperelastic skin. Real human skin is only a few millimeters thick. It can stretch and slide over underlying body structures such as muscles, bones, and tendons, revealing rich details of a moving character. Simulating such skin is challenging because it is in close contact with the body and shares its geometry. Despite major advances in simulating elastodynamics of cloth and soft bodies for computer graphics, such methods are difficult to use for simulating thin skin due to the need to deal with non-conforming meshes, collision detection, and contact response. We propose a novel Eulerian representation of skin that avoids all the difficulties of constraining the skin to lie on the body surface by working directly on the surface itself. Skin is modeled as a 2D hyperelastic membrane with arbitrary topology, which makes it easy to cover an entire character or object. Unlike most Eulerian simulations, we do not require a regular grid and can use triangular meshes to model body and skin geometry. The method is easy to implement, and can use low resolution meshes to animate high resolution details stored in texture-like maps. Skin movement is driven by the animation of body shape prescribed by an artist or by another simulation, and so it can be easily added as a post-processing stage to an existing animation pipeline. We provide several examples simulating human and animal skin, and skin-tight clothes.

Thin Skin Elastodynamics

A Hybrid Lagrangian-Eulerian Formulation for Bubble Generation and Dynamics

Saket Patkar, Mridul Aanjaneya, Dimitriy Karpman, Ronald Fedkiw

We present a hybrid Lagrangian-Eulerian framework for simulating both small and large scale bubble dynamics, where the bubbles can grow or shrink in volume as dictated by pressure forces in the surrounding fluid. Small under-resolved bubbles are evolved using Lagrangian particles that are monolithically two-way coupled to the surrounding flow in a manner that closely approximates the analytic bubble oscillation frequency while converging to the analytic volume as predicted by the well-known Rayleigh-Plesset equation. We present a novel scheme for interconverting between these under-resolved Lagrangian bubbles and larger well-resolved bubbles that are modeled with a traditional Eulerian level set approach. We also present a novel seeding mechanism to realistically generate bubbles when simulating fluid structure interaction with complex objects such as ship propellers. Moreover, our framework for bubble generation is general enough to be incorporated into all grid-based as well as particle-based fluid simulation methods.

A Hybrid Lagrangian-Eulerian Formulation for Bubble Generation and Dynamics