Journal for Geometry and Graphics 22 (2018), No. 1, 115--127
Copyright Heldermann Verlag 2018
Data-Driven Approach for Enhancing Splashing Effects of Liquid Simulations
Takashi Kanai
Graduate School of Arts and Sciences, University of Tokyo, 3-8-1 Komaba, Meguro-ku, Tokyo 153-8902, Japan
kanait@acm.org
Mengyuan Wan
Graduate School of Arts and Sciences, University of Tokyo, 3-8-1 Komaba, Meguro-ku, Tokyo 153-8902, Japan
Fumiko Enomoto
Graduate School of Arts and Sciences, University of Tokyo, 3-8-1 Komaba, Meguro-ku, Tokyo 153-8902, Japan
In liquid simulation, splashing is a phenomenon where small particles scatter in the movement of liquids. Fine liquid simulation on the high-resolution grid is indispensable for obtaining detailed splashing effects. However, simulation costs increase and become too high to bear with increasing resolution. Various data-driven methods based on the idea of reusing pre-computed data have been attempted to shorten simulation time, but cannot be easily applied to liquid simulation directly. In this paper, we propose a data-driven method for enhancing visual effects such as splashing in liquid simulation. Our method successfully applies the data-driven approach to the FLIP method which is suitable for liquid simulation. Moreover, velocity fields can be up-sampled directly in run-time simulation. This enables enhancement of the splashing effect with the change in the topology of fluid surfaces. Our method is able to apply liquid motions that are different from pre-computed scenes. This realizes more flexible enhancements in run-time simulation. We also discuss database creation and usage, especially whether multiple databases of vector fields improve the quality of enhancement or not.
Keywords: Data-driven method, Liquid simulation, Splashing, FLIP, SVD.
MSC: 68U05; 65D15
[ Fulltext-pdf (5580 KB)]