Numerical and Geometric Algorithms for Virtual Practical Effects

点击翻译按钮获取中文摘要

In our 2010 Science article, we coined the term 'virtual practical effects' to describe an emerging methodology for technical artists to create digital effects. Supported by physics-based simulation technology, artists can use their knowledge of (and natural intuition for) the real world to virtually build 'physical' mechanisms that create the desired effects. Virtual practical effects has already shown much greater productivity compared to laboriously modelling every phenomenon at a lower level. For example, synchronized crashing ocean waves in 'Avatar' were produced by virtual wave generators modelled after the real thing, running in a physical simulation of the water we wrote. This approach is still in its infancy: my research program tackles the algorithmic challenges still standing in the way of a visual effects revolution. Part of the research will be dedicated to core low level algorithms, such as efficiently solving the linear and nonlinear systems arising in (multi-)physics simulations, or accurately and robustly tracking a detailed surface as it evolves through time. The other part of the research is concerned with more efficient or more capable numerical models of physical phenomena, like deep ocean waves interacting with boats, solid objects fracturing, or the deformation of thin film bubbles. The visual effects industry is important and rapidly growing within Canada, and Canada already leads the world in developing 3D animation software (e.g. Maya, Houdini). This program will build on my past research success in these industries, fuelling further growth. In the longer term, it also represents an important step in bringing visual effects capability to anyone with artistic talent and a story to tell - but not the time or inclination to wrestle with low-level technical effects work. Ultimately, the same goals of efficient fidelity to the real world and robustness in the hands of non-numerically-inclined users apply to much broader problems, in particular bringing numerical simulation prototyping to the do-it-yourself/maker revolution.