Early tests with the multitouch table hardware
Pfffeww ! I finally completed the construction of my mini multitouch project this weekend. In total, I probably spent 5 or 6 hours actually building, the rest of the time (spread over a few weeks since the beginning august) was spent scavenging and buying materials, parts and tools. The most frustrating part was that I had access to a very limited set of power tools (basically only a drill and a dremel). I had to do everything else using a hacksaw, which sucks for cutting wood ! Hopefully my uncle let me use his table saw when I needed to cut the acrylic sheet at a 45 degrees angle. The funiest part was when I went to get my ghetto IR filter (basically an unexposed film negative). I bought a brand new film and told the clerk to develop the film right away. You should’ve seen the look on his face ?! All I wanted is to get the film processed so that the unexposed parts turn black, which by the way block visible light, but not infrared ![]()

Anyhow, now that the hardware is essentially completed. It’s probably one of the very few wooden firewire peripherals you’ll ever see
I did some spray painting with the stencil I carved last week, so now it has the Creative Commons logo on the side.
Here are some pictures and a quick video I generated by some code I hacked with Processing.
Now I need to code a whole bunch of software. I’ll probably start by porting some of the code from Processing to a clean Java project. Then I’ll somehow need to hack some code for fixing that heavy barrel distortion caused by the use of a 1.9mm fisheye lens.
I’ll keep you guys updated.
That looks awesome! Any chance you could post the Processing code? I’ve never used Processing but I’m curious to see what the code for that is like.
Is this video really from your device?
Wow that’s quite impressive for a start, to say the least!!
bravo.c est un bon depart. j ai hate de voir ou
ca va deboucher.
Yes, it is indeed a video I extracted from the Processing output. I used the BlobDetection library for Processing. I’ll need to port it to Java.
Nice stuff. Maybe it will be easier than you tough
That camera lens produce a huge distortion, you’ll have to correct it to gain precision
Your multi-touch demo is very cool! You’ll have to show me in person before I leave!
I ported BlobDetection library to C++, so porting it to Java should even simpler
Nice video, keep us posted about heavy barrel distortion fix
Wow!
que dire de plus! Vraiment impressionant Francois!
Another idea on the same subject :
http://www.thethinkingblog.com/2007/09/how-to-make-your-own-multi-touch-pad.html
Algorithms for correcting barrel distortion are all over the place, it is a well known problem in machine vision. There are plenty of parallelized implementations (I’ll probably use java.utili.concurrent facilities)
The box needs to be redesigned to be more appealing…
In all, it is very impressing. Good luck for the barrel distortion correction algorithm!
[...] times over this problem of abstracting multitouch gestures, especially during the time I worked on my infrared LED-based mini multitouch table project last summer. I still wasn’t satisfied with my design until I saw the GestureMatch sample code provided by [...]