1. Prepare you machine You need to have puredata and Gem installed on your system. And you need to know where they are… You also need a compiling toolchain, e.g. GCC on Linux/Mac or Microsoft Visual C++ on Windows. 2. Get the sources : Download the pix_opencv sources from Puredata external repository : svn co svn://svn.code.sf.net/p/pure-data/svn/trunk/externals/pix_opencv or if you don't have svnl, can get a "snapshot",i.e. a zip file with the current code, here : https://sourceforge.net/p/pure-data/svn/HEAD/tree/trunk/externals/pix_opencv/ 3. Get opencv : Ubuntu/Linux The easiest way to get it is to install it from the repository : sudo apt-get install libopencv-* Mac OSX : The easiest way is to get from some packaging repository like Macports, fink or homebrew. with fink : sudo fink install opencv-dev with macport : sudo port install opencv Windows : Download the binary release on opencv.org You can follow the quickstart guide on opencv.org to setup environment variables. 4. Build the Sources : Ubuntu/Linux & MacOSX Go to the fresh created folder, i.e. « pix_opencv » and build the things : cd pix_opencv make If you see something like « can’t find m_pd.h », you need to tweak pd’s path with this command : make CFLAGS=-I/path/to/pd If you see something like « can’t find Base/GemPixObj.h », you need to tweak Gem’s path with this command : make CFLAGS=-I/path/to/gem You can combine options like : make CFLAGS=-I/path/to/pd CFLAGS=-I/path/to/gem Windows You need Visual C++ (at least the free express edition) to build pix_opencv on Windows. There is a solution in build/vs2010 folder. It was made with Visual C++ Express 2010 and surely need some tweaks before producing any DLL. You need to change the include and the library paths. The include must contain the following : - puredata include (pd-extended) or src (vanilla) folder : path\to\pd\src or path\to\pd-extended\include - Gem include folder : path\to\Gem - OpenCV include path, wich is ususally $(OPENCV_DIR)\..\..\include if you setup OPENCV_DIR according to Windows quickstart guide on opencv.org.