fluid simulation

Accelerating Eulerian Fluid Simulation with Convolutional Networks  Smoke
fluid simulation 3d ago 
Accelerating Eulerian Fluid Simulation with Convolutional Networks  Archway
fluid simulation 3d ago 
Accelerating Eulerian Fluid Simulation with Convolutional Networks  Stanford Bunny
fluid simulation 3d ago 
Fluid Particles
fluid simulation 1mo ago 
Two Minute Papers  Narrow Band Liquid Simulations
fluid simulation 2mo ago 
A Dimensionreduced Pressure Solver for Liquid Simulations
fluid simulation 3mo ago 
Matching Fluid Simulation Elements to Surface Geometry and Topology
fluid simulation 5mo ago 
Realtime 2D Fluid Simulation
fluid simulation 5mo ago 
Sound Reactive Fluid Sim Test  Cantina Creative
fluid simulation 6mo ago 
Datadriven Fluid Simulations using Regression Forests
fluid simulation 8mo ago 
Realistic Fluid Simulations
fluid simulation 9mo ago 
Two Minute Papers  Adaptive Fluid Simulations
fluid simulation 10mo ago 
Fluid simulation  Pouring water
fluid simulation 10mo ago 
Fluid SimulationFinale
fluid simulation 11mo ago 
Blender Fluid Simulation 1
fluid simulation 1y ago 
Simple 4K Blender Fluid Simulation
fluid simulation 1y ago 
HTML5 Fluid Simulation in WebGL
fluid simulation 1y ago 
Using Phoenix FD to Create Liquid Simulations  Wine Part 2
fluid simulation 2y ago 
Using Phoenix FD to Create Liquid Simulations  Wine Part 1
fluid simulation 2y ago 
Using Phoenix FD to Create Liquid Simulations  Chocolate
fluid simulation 2y ago 
High viscosity fluid simulation  RealFlow, 3ds Max, Nuke, After Effects  FULL HD
fluid simulation 2y ago 
ofxFlowtools addon for openframeworks
fluid simulation 2y ago 
Swoosh Visualiser Fluid Simulation
fluid simulation 2y ago 
Lil Stormy Cloud the process h264 UQ
fluid simulation 2y ago
Tags
Description
Jonathan Tompson, Kristofer Schlachter, Pablo Sprechmann, Ken Perlin Realtime simulation of fluid and smoke is a long standing problem in computer graphics, where stateoftheart approaches require large compute resources, making realtime applications often impractical. In this work, we propose a datadriven approach that leverages the approximation power of deeplearning methods with the precision of standard fluid solvers to obtain both fast and highly realistic simulations. The proposed method solves the incompressible Euler equations following the standard operator splitting method in which a large, often illcondition linear system must be solved. We propose replacement of this system by learning a Convolutional Network (ConvNet) from a training set of simulations using a semisupervised learning method to minimize longterm velocity divergence. ConvNets are amenable to efficient GPU implementations and, unlike exact iterative solvers, have fixed computational complexity and latency. The proposed hybrid approach restricts the learning task to a linear projection without modeling the well understood advection and body forces. We present realtime 2D and 3D simulations of fluids and smoke; the obtained results are realistic and show good generalization properties to unseen geometry. http://cims.nyu.edu/~schlacht/CNNFluids.html