Skip to content

Latest commit

 

History

History
145 lines (94 loc) · 5 KB

README.md

File metadata and controls

145 lines (94 loc) · 5 KB
Raw

Numerical Geometry Project: a fast implementation of the Delaunay Triangulation

  • Author: Jérome Eertmans (13551600)
  • Course: Numerical Geometries (LMECA2710)
  • Algorithm: Delaunay triangulation

How to use this project

If you want to use the project, you need to first compile it:

mkdir build     # only required once
cd build
cmake ..        # only required once
make
cd ..

Alternatively, you can use ./buid.sh if you are on a UNIX machine (please first create the directory and instantiate the cmake).

Then you can execute the program. A variety of parameters is available, please use -h flag to see them.

./build/bin/lmeca2710_project [-param value] ...

However, I highly recommend you to use my Python wrapper, as it adds very nice sound effects!

What you need:

  • A Python3 installation
  • 2 small Python packages (use: pip3 install -r requirements.txt)
  • 3 other Python packages if you want to use python3 benchmark.py (included in requirements.txt)

Then, you can use the produce the exact same results + some bonuses:

python3 main.py [-param value] ...

You can also modify the music theme in main.py.

Warning: key bindings assume you have an AZERTY layout. If your keyboard layout differs, please transpose the keys accordingly. Example: pressing Q on a QWERTY layout will be understood as the key A.

Drawing images with triangulation

Some funny images can be generated by triangulating very specific sets of points. Please have a look into the input folder.

The drawings data files were kindly generated by Pierre Veldeman using some public image found on the internet.

I edited some files manually to show you what you can do with this program.

 

Voronoi diagrams

From a given Delaunay triangulation, it is trivial to construct the equivalent Voronoi diagram. Such a possibility is provided by the program and is computed in a linear time complexity.

YOUPIDOU mode

If you happen know the Pr. Vincent Legat, then you should for sure have a look to the YOUPIDOU mode of execution using the -y flag with Python.

Sound files have been kindly provided by the Youpidapp authors. If you happen to have an Android phone, make sure to try their app!

All these sounds files are the property of Vincent Legat and were obtained from his Youtube channel.

Made to be fast

Event though the Fortune algorithm can be used to generate Voronoi diagrams (and thus Delaunay Triangluation) faster than the divide and conquer approach used here, this algorithm is also much harder to implement.

Here, a small benchmark (on the same computer) shows that a good implementation of the divide and conquer solution can be faster (more than 2 times!) than the Fortune's algorithm, implemented by two friends for the same course.

Contribute

If you find any error or wish to add any add-on on my code, feel free to contact me via Github issues :)

Known issues

Memory issues

In the GUI, using the SHIFT + KEY (i.e. fast repeat of KEY) can cause memory problems if used for too long.

This is the case when Voronoi and pretty drawings are activated, and that you try to quickly delete a lot of points.

After having a look at memory leaks with Valgrind, it seems like the memory problems may come from the BOV library, a library provided for the course. My code, in its own, is Memcheck clean as it does not leak anything when used with -d flag.

ALSA lib pcm.c:8526:(snd_pcm_recover) underrun occurred

You can ignore this error, it just happens because you closed the application when a new sound file was just starting to be played.

Sound bibliography

All the musics are Copyright Free songs. If you wan to reuse them, please make sure to keep it as a private usage.

Sound effects: http://soundbible.com

Synthwave theme: https://www.youtube.com/watch?v=onEyXlpdY3U

Chill theme: https://incompetech.com/music/royalty-free/index.html?isrc=USUAN1100326

Guides

This folder is the canvas for your upcoming project. It should contain:

  • this file (README.md)
  • the description of the structure of the program in CMakeLists.txt
  • a src directory containing the the source code of your program
  • a doc directory containing more documentation
  • a deps directory containing the BOV library

See doc/COMPILING.md for a step by step tutorial on how to build the program.

See doc/tutorial.md for a step by step tutorial on how to use the BOV library.

See deps/BOV/include/BOV.h for help on the BOV library functions.

See deps/BOV/examples/ for more examples using the BOV library