Unified GPU Framework for Simulating Wave Turbulence, Diffusion, and Wrinkling in Fluid-Cloth Interaction

This paper proposes a GPU-based simulation framework to efficiently represent fine-grained physical phenomena, such as turbulence, diffusion, and wrinkling, that occur in fluid-cloth interactions. Unlike previous methods that rely solely on momentum conservation, this framework captures the intricate interactions between fluids and fabrics. The FLIP approach is used to simulate more detailed fluid flows than traditional particle-based fluid techniques, while the PBD method captures the fundamental physical properties of fabrics. To efficiently handle large-scale details like turbulence, diffusion, and wrinkling, the framework utilizes GPU parallelization techniques, including spatial hashing, graph coloring, and memory optimization. The result is a significant performance improvement, with tens of times faster simulation times compared to CPU methods, enabling the rapid representation of fine-grained visual features.