2020/08/13 · IST Austria
IST Austria scientists present three novel techniques to simulate nature’s complexity
From large waterfalls to tiny soap bubbles – nature is both beautiful and complex. Simulations of its phenomena are essential to scientific discoveries, modern engineering, and the digital arts. While supercomputers are usually necessary, Professor Christopher Wojtan and his team overcome this complexity by combining physics, mathematics, and algorithmic optimizations from computer science.
Their research achieves some of the world’s fastest and most detailed simulations by understanding the underlying mathematical models and inventing computational techniques. At the Siggraph conference, the world’s largest conference on computer graphics and animation starting on August 17th 2020, scientists from the Wojtan group now present three discoveries.
Waves on top of waves
Waves are every traveler’s dream, but every 3D animator’s nightmare. While water simulation has become incredibly realistic in the past decade, its immense computing power leads to limitations, usually shown on the water surface.
Now, Tomáš Skřivan presents a new method to enhance a fluid surface’s details in a physically plausible manner, with little computational expense. His method takes an existing simulation and simulates many detailed water waves on top of it. The technique produces ripples with dispersive wave-like behaviors customized to the underlying fluid simulation.
For more information, check out the Wave Curves project page.
Yarn and thread
One wrong stitch decides the fate of a pullover or sock. How it feels, stretches, and bends only depends on the pattern and quality of stitches. While these are seemingly simple objects, their stiches and patterns are highly interesting from a mathematical point of view. For simulations, you could simulate each interacting thread, but this would take immense computation power. Another approach is to treat fibers as an averaged effective material and not individually. These simulations are more efficient, but finding a model that matches the original cloth is complicated.
However, Georg Sperl created a clever new strategy for finding realistic material models for fabrics. The researcher precomputes hundreds of simulations of yarns interacting with each other, automatically creates a computer model which reproduces these effects, and incorporates it into a cloth simulator.With this technique, the scientists derive a material’s properties directly from its geometry. No real-world experiments nor measurements are needed. With this, the research group created the first use of numerical homogenization for animating woven and knitted fabrics. Their technique is capable of reproducing common textile phenomena such as anisotropy, area preservation, and curling.
For more information, check out the homogenized yarn project page.
Soap films, bubbles, and foams are fascinating and beautiful in their geometry, dynamics, and color. Continuous changes in the soap film lead to the swirling patterns seen in the light. Previous animations only focused on the bubble’s surface and shape but neglected its thickness. However, these oversimplifications of previous methods prevent the appearance of characteristic bubble phenomena like vortices, ripple patterns, gravity-dependent thickness variation, and even bursting.
Sadashige Ishida and Peter Synak proposed to include a dynamic thickness of the soap film within bubble simulations. With mathematical equations, they described the shifting soap film and the deformation of soap bubbles. Their simulations enhance state-of-the-art bubble simulations with additional effects caused by convection, rippling, draining, and thin-film evaporation.
For more information, check out the evolving soap film project page.