View Installing Touchlib on Mac OS X
1. Install Apple Developer Tools. This package is included as an optional install on your original OS X installation disk. Alternatively, you can download the new version of Xcode here.
2. Install Darwinports. Darwinports offers more current software packages than Fink. Further information is available here.
3. Install the required packages using Darwinports (instructions on each linked page, or use Porticus for a GUI):
libjpeg
libtiff
libpng
cmake
fftw
opencv (If this doesn’t work, go to step 4. This has worked successfully with Porticus on both a G5 Tiger and an Intel Leopard)
If all of those install without errors skip to step 5. If you can’t get opencv to install, go to step 4 for instructions on building it yourself.
[Note: Python comes pre-installed on OS X Tiger]
4. Install OpenCV (only if you couldn’t get it through darwinports).
At the Terminal type:
cvs -d:pserver:anonymous@opencvlibrary.cvs.sourceforge.net:/cvsroot/opencvlibrary login
It will ask for a password. Just hit enter. Now, copy this in:
cvs -z3 -d:pserver:anonymous@opencvlibrary.cvs.sourceforge.net:/cvsroot/opencvlibrary co -P opencv
This will download the source code into a directory in your home folder. It is about 100MB and may take a while depending on your connection speed. Once this is done, navigate to the new directory in Terminal:
cd opencv
Now, we finally get to configure the source code. It would be wise to make a build folder by typing in the following:
mkdir build; cd build
The new versions of OpenCV will automatically compile Universal Binary. Configure OpenCV using the packages you just installed from Darwinports:
../configure CPPFLAGS="-I/opt/local/include" LDFLAGS="-L/opt/local/lib"
After everything is configured, install it using the command:
make
sudo make install
To double check that everything is working type:
make check
If there are no serious errors, everything OpenCV should be working.
5. Install OSCPack
Download oscpack here.
Replace their Makefile with this Makefile File:OscpackMakefileForMacOSX.txt (modified for OS X).
In the oscpack directory, make and install it:
make
sudo make install
6. Install SVN
Download svn here.
Note that you will need to add /usr/local/bin to either your account’s PATH, or the system level PATH if it has not already been altered for some other application. To modify to the path for all users, modify /etc/profile (for sh, ksh, bash, etc.) or /etc/csh.login (for csh, tcsh, etc).
7. Install Touchlib
Create a driectory for Touchlib and download the latest copy.
mkdir ~/Desktop/Touchlib/
cd ~/Desktop/Touchlib/
svn co http://touchlib.googlecode.com/svn/trunk/ multitouch
=============
FIXME::This next section should be resolved in the SVN repository ASAP
There are a couple small touchlib code changes needed in version 47
In the file ./src/CBlobTracker.cpp
change line 704 734 from:
if((i == (numcheck-1) && extraIDs > 0) || current[start].error > reject_distance_threshold)
to
if((i == (numcheck-1) && extraIDs > 0))// || current[start].error > reject_distance_threshold)
Next, in the file ./demos/smoke/fluid2d.h
change line 15 19 from:
#include <rfftw.h>
to
#include "fluid2d.h"
=============
Now navigate to the source code folder and compile:
cd multitouch
cmake .
make
There shouldn’t be any errors.
So, now you can test a couple programs.
cd src/
./sampleapp
or
./configapp
You can also try out the demo apps:
cd ./demos/smoke/
./smoke
and
cd ./demos/osc/
./osc
cd ./demos/
./osc
Pong will not work on Linux or OS X because it relies on some Windows libraries.
As you will probably see, the program will come up—you might even see yourself in the camera, then it will crash with some memory issues, bus errors or segmentation faults.
I have yet to check into these errors, but hope to soo. Good luck and as this is a wiki, please make changes for the sake of clarity/efficiency.
Christopher
References:
some info copied and pasted from other guides on this site as well as the OpenCV wiki.
*Please excuse the formatting neglect on this first revision.
**These instructions were developed on an Intel-based Macbook Pro, running OS X 10.4.9. While the instrctions should also work on older PPC machines Intel’s OpenCV library and the associated Integreated Performance Primitives were originally designed to take full advantage of Intel Chips.
