libWetCloth is an open source project for the physical simulation of liquid and wet cloth or yarns. It is cross-platform (Mac OS X, Linux, Windows, and more), and licensed under the Mozilla Public License v. 2.0.
We would like to hear from you if you appreciate this work.
It is the original implementation of paper A Multi-Scale Model for Simulating Liquid-Fabric Interactions ( refer to our project page for more details: http://libWetCloth.info, or http://www.cs.columbia.edu/cg/wetcloth/ ). This code base contains the following parts:
- A liquid simulator implementing the affine-particle-in-cell method.
- A cloth simulator implementing the elastic thin shell model.
- A yarn simulator implementing the discrete viscous thread model.
- A cloth/yarn collision handler based on anisotropic elastoplasticity, discretized with augmented, moving least sqaures material point method (AMLS-MPM).
- A two-way coupling method based on mixture theory, between the cloth, yarn and liquid, handling dragging, buoyancy, capturing and dripping effect.
libWetCloth depends on following libraries (some of them have been included in the code base, marked with asterisk):
- Eigen* (http://eigen.tuxfamily.org/)
- RapidXML* (http://rapidxml.sourceforge.net)
- tclap* (http://tclap.sourceforge.net)
- libIGL* (https://github.com/libigl/libigl)
- AntTweakBar (http://anttweakbar.sourceforge.net/doc/)
- Intel TBB (https://www.threadingbuildingblocks.org)
- GLUT (https://www.opengl.org/resources/libraries/glut/)
- libPNG (https://libpng.sourceforge.io/)
- zlib (https://www.zlib.net/)
On Mac OS X or Linux-based systems, most of the dependencies are either included, or can be easily installed with Homebrew (https://brew.sh) or the APT package handling utility. For Intel TBB, you may download and install from the link provided above, or from Intel website (https://software.intel.com/en-us/parallel-studio-xe/choose-download).
On Windows you may need manually download and compile some of them (e.g. AntTweakBar, TBB, libPNG). For libPNG and zlib, please make sure the pnglibconf.h and zconf.h have been copied to their "include" directories (i.e., the directories containing png.h and zlib.h, correspondingly). For more details, please refer to the compilation section below.
libWetCloth has been tested with Clang (under Mac OS X), GCC 4.8+ (under Linux), and Microsoft Visual Studio (under Windows 10).
To compile libWetCloth, you'll need CMake on Mac OS X or Linux, or CMake-GUI (https://cmake.org) on Windows.
On Mac OS X or Linux:
- make a directory, say, build, with mkdir build, enter the build directory, type cmake ..
- Optionally you can adjust the options with ccmake .. In some cases there can be some packages that cmake cannot find. You need to manually specify their paths through ccmake then.
- type make to compile the code. For speeding up the compilation process you may use make -j.
On Windows:
- open CMake-GUI, enter the correct directory for source code and build. Then click Configure, choose your installed version of the Microsoft Visual Studio.
- after configuration you may find several libraries not found (with notifications of errors), check the Advanced box and specify those missing header path and libraries manually. For example, if Eigen is missing, then please specify the EIGEN3_INCLUDE_DIR to the path of directory we provided. For the ones we have not provided, you need to download and compile them before compiling libWetCloth, and then specify the missing directories to the path containing your headers or compiled libraries. Please make sure you have picked the libraries corresponding to the architecture you have selected (say, 32-bit libraries for x86, and 64-bit libraries for x64).
- click generate after fixing all missing variables to generate your Visual Studio solution.
- open the Visual Studio solution and compile the code.
- before running the demo, all the compiled dynamic linking libraries (DLLs) for your dependencies should be accessible from your PATH environment variable that can be changed in system settings, or you may simply copy them into your System32 (x64) or SysWOW64 (x86) directories.
To run the demo of libWetCloth, you may simply use the command line argument -s [scene_file] to specify the scene to be loaded. For example, you may type
./libWetCloth -s assets/general_examples/splash_cloth_small.xml
to run the simulation of the scene containing a water ball splashes on a small cloth.
All the parameters can be modified offline in the scene description XML files. Some can be changed online in the user interface provided by the demo program.
USAGE:
./libWetCloth -s [-i ] [-o ] [-g ] [-d ] [-p ] [--] [--version] [-h]
Where: -s , --scene (required) Simulation to run; an xml scene file
-i , --inputfile Binary file to load simulation pos from
-o , --outputfile Binary file to save simulation state to
-g , --generate Generate PNG if 1, not if 0
-d , --display Run the simulation with display enabled if 1, without if 0
-p , --paused Begin the simulation paused if 1, running if 0
--, --ignore_rest Ignores the rest of the labeled arguments following this flag.
--version Displays version information and exits.
-h, --help Displays usage information and exits.
The Houdini projects are also provided in the "houdini" folder, which are used for surface reconstruction and rendering purposes. Our simulator can generate data that can be read back by the Python script in our Houdini projects.
You may run the demo with the "-o" option, under the folder containing Houdini projects (by default, [project source]/houdini/). For example, you may type
../build/libWetCloth/libWetCloth -s ../assets/parameter_tests/pore_test_mid.xml -o 20
under the folder containing Houdini projects to generate data per 20 time steps (for this example we use 0.0002s for the simulation time step, and 0.004s for the rendering time step. Hence 20 time steps is taken for the data generation). The simulation code will create a folder with the name of the scene file under this folder ([project source]/houdini/pore_test_mid for this example), which can be read back by the Houdini project with the same name.
After some data generated, you may open the corresponding Houdini project to watch and operate on them. We use the nodes with suffix "bake" to indicate the usage of baking. For example in the "fluids_bake" node, the fluid particles will be read and used to reconstruct a polygonal liquid surface, which will then be stored as Houdini geometry files.
The baked geometries are then read back by the nodes with prefix "geo", which can be used for rendering.
You may check the Mantra nodes in the output classifier for the rendering settings. You can also connect these nodes to HQueue nodes for distributed rendering on a render farm.
Houdini 16.5 is used for our case. You may need an equivalent or higher version to open the project files. For more information and tutorials of Houdini, please visit the SideFX website (https://www.sidefx.com/).
Please contact the author ([email protected]) for questions and bug report, or consulting for the usage of this code base.
@article{Fei:2018:MMS:3197517.3201392
author = {Fei, Yun (Raymond) and Batty, Christopher and Grinspun, Eitan and Zheng, Changxi},
title = {A Multi-scale Model for Simulating Liquid-fabric Interactions},
journal = {ACM Trans. Graph.},
issue_date = {Aug 2018},
volume = {37},
number = {4},
month = aug,
year = {2018},
pages = {51:1--51:16},
articleno = {51},
numpages = {16},
url = {http://doi.acm.org/10.1145/3197517.3201392},
doi = {10.1145/3197517.3201392},
acmid = {3201392},
publisher = {ACM},
address = {New York, NY, USA},
keywords = {fluid dynamics, particle-in-cell, mixture theory, two-way coupling, wet cloth},
}