135 lines
5.1 KiB
Markdown
135 lines
5.1 KiB
Markdown
|
# QuadriFlow: A Scalable and Robust Method for Quadrangulation
|
||
|
|
||
|
Source code for the paper:
|
||
|
|
||
|
Jingwei Huang, Yichao Zhou, Matthias Niessner, Jonathan Shewchuk and Leonidas Guibas. [**QuadriFlow: A Scalable and Robust Method for Quadrangulation**](http://stanford.edu/~jingweih/papers/quadriflow/quadriflow.pdf), The Eurographics Symposium on Geometry Processing (SGP) 2018.
|
||
|
|
||
|
<!-- ## Processing Result -->
|
||
|
![QuadriFlow Results](https://github.com/hjwdzh/quadriflow/raw/master/img/result.jpg)
|
||
|
|
||
|
## WebGL Application
|
||
|
Our 3D WebGL Apps for QuadriFlow are online! Without any installation, you are able to
|
||
|
* [**Compare**](https://yichaozhou.com/publication/1805quadriflow/#demo) QuadriFlow with previous methods;
|
||
|
* [**Quadrangulate**](https://yichaozhou.com/publication/1805quadriflow/#tool) your own meshes and
|
||
|
download the result!
|
||
|
|
||
|
## Desktop Software
|
||
|
The software supports cmake build for Linux/Mac/Windows systems. For linux and mac users, run **`sh demo.sh`** to build and try the QuadriFlow example, which converts `examples/Gargoyle_input.obj` to `examples/Gargoyle_quadriflow.obj`.
|
||
|
|
||
|
### Install
|
||
|
|
||
|
```
|
||
|
git clone git://github.com/hjwdzh/quadriflow
|
||
|
cd quadriflow
|
||
|
mkdir build
|
||
|
cd build
|
||
|
cmake .. -DCMAKE_BUILD_TYPE=release
|
||
|
make -j
|
||
|
```
|
||
|
|
||
|
### QuadriFlow Software
|
||
|
|
||
|
We take a manifold triangle mesh `input.obj` and generate a manifold quad mesh `output.obj`. The face number increases linearly with the resolution controled by the user.
|
||
|
|
||
|
```
|
||
|
./quadriflow -i input.obj -o output.obj -f [resolution]
|
||
|
```
|
||
|
|
||
|
Here, the resolution is the desired number of faces in the quad mesh.
|
||
|
|
||
|
## Advanced Functions
|
||
|
|
||
|
### Min-cost Flow
|
||
|
By default, `quadriflow` uses the Boykov maximum flow solver from boost because it is faster. To
|
||
|
enable the adaptive network simplex minimum-cost flow solver, you can enable the `-mcf` option:
|
||
|
|
||
|
```
|
||
|
./quadriflow -mcf -i input.obj -o output.obj -f [resolution]
|
||
|
```
|
||
|
|
||
|
### Sharp Preserving
|
||
|
By default, `quadriflow` does not explicitly detect and preserve the sharp edges in the model. To
|
||
|
enable this feature, uses
|
||
|
|
||
|
```
|
||
|
./quadriflow -sharp -i input.obj -o output.obj -f [resolution]
|
||
|
```
|
||
|
|
||
|
### SAT Flip Removal (Unix Only)
|
||
|
By default, `quadriflow` does not use the SAT solver to remove the flips in the integer offsets
|
||
|
map. To remove the flips and guarantee a watertight result mesh, you can enable the SAT solver.
|
||
|
First, make sure that `minisat` and `timeout` is properly installed under your `${PATH}`. The
|
||
|
former can be done by building `3rd/MapleCOMSPS_LRB/CMakeLists.txt` and copying `minisat` to `/usr/bin`.
|
||
|
In addition, `timeout` is included in coreutils. If you are using Mac, you can install it using
|
||
|
homebrew:
|
||
|
```
|
||
|
brew install coreutils
|
||
|
export PATH="/usr/local/opt/coreutils/libexec/gnubin:$PATH"
|
||
|
```
|
||
|
|
||
|
You can verify if those binaries are properly installed by executing
|
||
|
```
|
||
|
which minisat
|
||
|
which timeout
|
||
|
```
|
||
|
|
||
|
After that, you can enable SAT flip removal procedure by executing
|
||
|
```
|
||
|
./quadriflow -sat -i input.obj -o output.obj -f [resolution]
|
||
|
```
|
||
|
|
||
|
When using the SAT flip removal, we also suggest you enabling the verbose logging to understand
|
||
|
what is going on. You can build quadriflow with the following options:
|
||
|
```
|
||
|
cmake .. -DCMAKE_BUILD_TYPE=release -DBUILD_LOG=ON
|
||
|
```
|
||
|
|
||
|
### GUROBI Support (For Benchmark Purpose)
|
||
|
|
||
|
To use the Gurobi integer programming to solve the integer offset problem, you can build QuadriFlow with
|
||
|
```
|
||
|
cmake .. -DCMAKE_BUILD_TYPE=release -DBUILD_GUROBI=ON -DBUILD_LOG=ON
|
||
|
```
|
||
|
This override other solvers and should only be used for benchmark purpose.
|
||
|
|
||
|
## External Dependencies
|
||
|
* Boost
|
||
|
* Eigen
|
||
|
* OpenMP (optional in CMake)
|
||
|
* TBB (optional in CMake)
|
||
|
* GUROBI (for benchmark purpose only)
|
||
|
|
||
|
## Licenses
|
||
|
|
||
|
QuadriFlow is released under [MIT License](LICENSE.txt).
|
||
|
For 3rd dependencies,
|
||
|
* Boost and Lemon are released under [Boost Software License](https://lemon.cs.elte.hu/trac/lemon/wiki/License)
|
||
|
* Most part of Eigen is released under [MPL2](https://www.mozilla.org/en-US/MPL/2.0/FAQ/)
|
||
|
* Sparse Cholesky Decomposition algorithms are released under LGPL
|
||
|
* To replace it using Sparse LU decomposition with a more permissive MPL2 license (a little slower), enable `BUILD_FREE_LICENSE` in CMake (e.g., `-DBUILD_FREE_LICENSE=ON`).
|
||
|
* `pcg32.h` is released under the Apache License, Version 2.0
|
||
|
* `parallel_stable_sort.h` is released under the MIT License
|
||
|
|
||
|
## Authors
|
||
|
- [Jingwei Huang](mailto:jingweih@stanford.edu)
|
||
|
- [Yichao Zhou](mailto:zyc@berkeley.edu)
|
||
|
|
||
|
© 2018 Jingwei Huang and Yichao Zhou All Rights Reserved
|
||
|
|
||
|
**IMPORTANT**: If you use this software please cite the following in any resulting publication:
|
||
|
```
|
||
|
@article {10.1111:cgf.13498,
|
||
|
journal = {Computer Graphics Forum},
|
||
|
title = {{QuadriFlow: A Scalable and Robust Method for Quadrangulation}},
|
||
|
author = {Huang, Jingwei and Zhou, Yichao and Niessner, Matthias and Shewchuk, Jonathan Richard and Guibas, Leonidas J.},
|
||
|
year = {2018},
|
||
|
publisher = {The Eurographics Association and John Wiley & Sons Ltd.},
|
||
|
ISSN = {1467-8659},
|
||
|
DOI = {10.1111/cgf.13498}
|
||
|
}
|
||
|
```
|
||
|
|
||
|
## Triangle Manifold
|
||
|
|
||
|
In case you are dealing with a triangle mesh that is not a manifold, we implemented the software that converts any triangle mesh to watertight manifold. Please visit https://github.com/hjwdzh/Manifold for details.
|