Skip to main content

Measurement-Based Modeling and Animation of Complex Mechanical Phenomena

Final Report Summary - ANIMETRICS (Measurement-Based Modeling and Animation of Complex Mechanical Phenomena)

The Animetrics project has investigated novel modeling methods for the animation of a broad set of mechanical effects with a visual output, covering cloth deformation, tissue biomechanics, contact, fracture, and fluid dynamics. The common thread of the project is to exploit data from real-world or precomputed examples to devise more efficient, versatile, and/or realistic models.

The major goals achieved can be grouped as:

1. Modeling nonlinear solid mechanics from examples.
Progress in solid simulation for computer animation has led to a multitude of deformation models, each with its own way of relating geometry, deformation, and forces. We have designed measurement and fitting methods that allow nonlinear models to be fit to the observed deformation of a particular object. Unlike standard engineering testing, our system measures complex 3D deformations of an object, not just one-dimensional force–displacement curves, so it works under a wider range of deformation conditions.

2. Simulation of cloth at the yarn level.
The large-scale mechanical behavior of cloth is determined by the mechanical properties of the yarns, the weave or knit pattern, and frictional contact between yarns. Using standard simulation methods that represent the individual yarns and yarn-yarn contact handling, the simulation of garments at realistic yarn densities was deemed intractable in the past. We introduce an efficient solution for simulating both woven and knitted cloth at the yarn level. Central to our solution is a novel representation of interlaced or stitched yarns based on yarn contacts and yarn sliding, which allows modeling yarn-yarn contact implicitly, avoiding yarn contact handling altogether. Combined with models for internal yarn forces and inter-yarn frictional contact, as well as a massively parallel solver, we are able to simulate garments with over a million yarn crossings.

3. Efficient modeling of nonlinear soft tissue mechanics.
Applications of grasping simulation in robotics, haptics, or virtual reality require thorough knowledge of the forces and deformations caused by contact interactions on the skin. To answer this need, we propose an efficient model to simulate the skin’s biomechanics under frictional contact. The key novelty in our approach is to model the extremely nonlinear elasticity of finger skin and flesh using deformation-limiting constraints, which are seamlessly combined with frictional contact in a standard solver. We show that our approach enables haptic simulation of rich and compelling deformations of the fingertip. We combine it with a novel constrained dynamics solver and optimization methods for the control of haptic devices to produce direct touch interactions on the fingers.

4. Interactive deformation of medical image data.
The standard pipeline for medical simulation of patient-specific image data requires careful segmentation of anatomical structures, complex meshing, and fine parameter tuning. In contrast to this pipeline, we present a lightweight method to interactively deform volumetric medical data with heterogeneous structural content. Our method rests on two major components: a massively parallel algorithm for the rasterization of deformable meshes, and a method to define a coarse deformable representation from the homogenization of a fine representation. Our homogenization method incorporates nonlinear shape functions and run-time treatment of accurate high-resolution boundary conditions.

5. Geometric fracture methods with example-based acceleration.
Previous methods for the animation of fracture are either physically-based, which allows the creation of crack patterns as a result of impact simulations, or geometry-based, which allows controllable editing of crack patterns by an artist. We propose a novel algorithm to simulate brittle fracture that improves the versatility of previous methods and their ability to adapt fracture patterns automatically to diverse collision scenarios and object properties. We cast brittle fracture as the computation of a high-dimensional Centroidal Voronoi Diagram, where the distribution of fracture fragments is guided by the deformation field of the fractured object.

6. Models of complex liquid surface motion based on convolution kernels.
We have introduced a radically novel formulation of surface wave models that accounts for the dispersion effects in common liquids. Previous methods addressed this problem in the frequency domain, but could not deal with nonlinearities introduced by boundary conditions. We have efficiently turned the problem into a spatial multi-scale formulation with handling of boundary conditions. To do this, we have introduced convolution kernels, referred to as 'dispersion kernels' that efficiently compute liquid surface evolution.