Fluid Simulation on Neural Flow Maps
arXiv:2312.14635 · doi:10.1145/3618392
Abstract
We introduce Neural Flow Maps, a novel simulation method bridging the emerging paradigm of implicit neural representations with fluid simulation based on the theory of flow maps, to achieve state-of-the-art simulation of inviscid fluid phenomena. We devise a novel hybrid neural field representation, Spatially Sparse Neural Fields (SSNF), which fuses small neural networks with a pyramid of overlapping, multi-resolution, and spatially sparse grids, to compactly represent long-term spatiotemporal velocity fields at high accuracy. With this neural velocity buffer in hand, we compute long-term, bidirectional flow maps and their Jacobians in a mechanistically symmetric manner, to facilitate drastic accuracy improvement over existing solutions. These long-range, bidirectional flow maps enable high advection accuracy with low dissipation, which in turn facilitates high-fidelity incompressible flow simulations that manifest intricate vortical structures. We demonstrate the efficacy of our neural fluid simulation in a variety of challenging simulation scenarios, including leapfrogging vortices, colliding vortices, vortex reconnections, as well as vortex generation from moving obstacles and density differences. Our examples show increased performance over existing methods in terms of energy conservation, visual complexity, adherence to experimental observations, and preservation of detailed vortical structures.
References in corpus (16)
- Instant Neural Graphics Primitives with a Multiresolution Hash Encoding
- Implicit Neural Representations with Periodic Activation Functions
- HyperNeRF: A Higher-Dimensional Representation for Topologically Varying Neural Radiance Fields
- Physics Informed Neural Fields for Smoke Reconstruction with Sparse Data
- ScalarFlow: A Large-Scale Volumetric Data Set of Real-world Scalar Transport Flows for Computer Animation and Machine Learning
- Lagrangian Neural Style Transfer for Fluids
- Direct Voxel Grid Optimization: Super-fast Convergence for Radiance Fields Reconstruction
- Learned Coarse Models for Efficient Turbulence Simulation
- 3D Neural Scene Representations for Visuomotor Control
- A Characteristic Mapping Method for the three-dimensional incompressible Euler equations
- Learning Object-Centric Neural Scattering Functions for Free-Viewpoint Relighting and Scene Composition
- PhySG: Inverse Rendering with Spherical Gaussians for Physics-based Material Editing and Relighting
- NeuralVDB: High-resolution Sparse Volume Representation using Hierarchical Neural Networks
- Neural UpFlow: A Scene Flow Learning Approach to Increase the Apparent Resolution of Particle-Based Liquids
- Fast Aquatic Swimmer Optimization with Differentiable Projective Dynamics and Neural Network Hydrodynamic Models
- Learning Vortex Dynamics for Fluid Inference and Prediction