Physically-based Flow Field Modulation And Multi-fluid Simulation

The development of technology and improvement of life quality have brought increasing requirements over reality and interactivity of the digital products. As an important part of the real world, various fluid phenomena play very important roles in creating realistic digital scenes. However, the complex physical mechanism of the real-world fluid phenomena makes fluid simulation need many computational resources and be time consuming. As a result, producing realistic and visually plausible results with limited computational resources and time in an e?cient manner remains an important topic in fluid simulation in computer graphics. Physically-based fluid simulation is able to allow people produce realistic and interactive fluid motions with a small number of manipulations, and it is thus welcomed in computer graphics and has received continuous attention.In physically-based simulation there are several important topics that require deep understanding of the complex fluid interaction mechanism. We propose physically-based fluid simulation methods fit for computer graphics application based on researches over these aspects, providing practical methods that can find merits in fluid control, expanding simulation applicability or fluid simulation acceleration. The main contributions include:1. We present a fluid control method based on flow field decomposition and modulation. The proposed method decomposes flow field and analyzes its intrinsic multiscale spatial frequency information. A novel concept of “flow style” is quantitatively defined during this process. Based on amplitude modulation using the above intrinsic information of flow field, the proposed method can e?ciently control the visual features of the flow at different scales. This method provides an effective solution to directly control visual appearance of the flow field, and provides an easy and practical way to handle the consistency control problem in fluid simulation.2. We propose a universal and stable multiple-fluid simulation method. The proposed method can capture a wide range of real-world multiple-fluid phenomena, including mixing/unmixing of miscible and immiscible fluids, diffusion effect and chemical reaction, etc. The new multiple-fluid scheme can be readily integrated into existing state-of-the-art single-fluid simulators, and the multiple-fluid simulation is easy to set up with GPU parallelization.3. We present a fast particle-based gas simulation method. By compensating density and force calculation of the visible gas particles in the simulation, the proposed method can completely remove the ambient gas particles and the related computational costs required in traditional particle-based gas simulation methods, and provide 10-400 times speed up according to different scene settings. A particle splitting and merging scheme is also proposed to enable e?cient simulation of complex scenes involving liquid-gas transition.