Lights. Camera. Action!

With everything in place, I performed the first true test of the multi-touch surface, using the Touchlib library.

The camera was still extremely sensitive to visible light, so much so that the blobs were not being detected correctly and there was a lot of interference. The proper way to filter out the visible light spectrum would be to use an IR band-pass filter with the right wavelength for the LEDs (in our case, 880nm) so that only the LEDs’ light can reach the camera (and nothing else, even in the infrared range). Instead of buying a filter, I decided to use the lazy-man’s visible light filter: exposed negative film. I placed three exposed negatives on top of the camera lens and the blobs were instantly clearer and the software could detect blobs and movement extremely accurately.

Once again in my dad’s shop and with his expert help, he and I constructed the frame for the table surface. This frame made transporting the surface and hooking it up so much easier—it was no longer a hassle to set up. Now it was finally time to start doing the actual multi-touch programming.

We soldered the 40 LEDs together (8 in each set connected in parallel, with the sets wired serially) and then performed our first test. It worked!

In my dad’s shop and with his help, we prepared the plexiglass for the LEDs:

Lights. Camera. Action.

Almost there! Now that the crazy semester is over, we’ll hopefully have a little bit more time to work on this. We received our camera from Unibrain and are very pleased with it so far. The documentation for the Unibrain Fire-i Board Camera, though not “clean” per se, is very helpful and gives you the impression that this device was designed for projects like this. Actually, that’s not too far from the truth as these cameras are designed for use in robotics.

We’ve done some small experimentation to determine the best way to filter visible light from the camera. Many blogs suggest that two fully-exposed negatives placed on top of the lens is a cheap man’s filter, but workable. We’ve found that it does increase contrast and makes blobs more recognizable, but still let’s most visible light through. There are other, more expensive solutions. The first is to use an IR pass filter to only let through a range of IR light (say 800-900 nm). Some suggest using an IR band-pass filter which may provide better results—it would allow only a specific wavelength of light which should be the wavelength of IR emitted by your IR LEDs (in our case, for our Osram SFH485 LEDs, we would get a IR band-filter of 880 nm).

Our acrylic is here also! It is 122×61 cm (48×24 in) and is 10 mm thick (we will be cutting it to a 4:3 ratio, approximately 61×46 cm or 24×18 in). We plan on machining the acrylic to place the LEDs directly into the edges, but first must test with different LED configurations to find the optimal number and placement for our setup.

Also, some great news for the open source multi-touch community: Google Summer of Code received over 42 applications for code projects related to the Natural User Interface Group, seven of which were accepted. The GSoC page contains all the accepted projects and descriptions, but we are very excited for one in particular: the Mac OS X OpenTouch Conversion. For now, we plan to pursue Linux as our main platform for development (and will post soon on how to configure the Unibrain Fire-i on Ubuntu), but the OpenTouch port to OS X would allow us to create multi-touch apps that use Core Animation, the native OS X graphics framework.

Lights. Camera. Action.

The first step towards our physical assembly for the FTIR table has been purchased—we have ordered a pack of Osram SFH485 IR LEDs. Creating a FTIR table requires a lot more to be done to the display surface, as opposed to a DI device where most of the extras are separate from the display.

For our display we are going to use a sheet of acrylic that is 24×18 inches (~ 61×46 cm) and placing the lights along the 24 in. sides. At this point, until we do some tests and more research, we are planning to use about 15 LEDs per side.

We ordered a pack of 100 LEDs from Digi-Key which should give us some extras to create another table or experiment later.

Once we have our our camera and acrylic we plan on doing some tests to see the best number of LEDs to use. From some blog posts we hear 1.5 centimeter apart, from others we hear that you should only need about 12 per side. We’ll see what works best when we get the equipment all set up.

Also, make sure to check out Jason Modisette’s great FTIR Screen Design Applet to help understand and visualize total internal reflection of the infrared light within the acrylic sheet.

Welcome to the uNUI Group! We are a small group who have decided to explore the realm of multi-touch interfaces as an extension of the NUI Group community. As we learn about the intricacies of building our own multi-touch devices, we will be documenting as much as we can on this website to help others through sharing of our experiences.

We are unsure of our future and the path we will find—it may revolve around building hardware and different devices, it may be primarily based in developing interface software and applications—we don’t know yet where our interests will lead us. Though, it is clear to us and many others that this is the future, that computers will relatively soon lose the cumbersome and disjointed interface of manipulating one pointer on an X-Y axis. The future of computer interaction will be multi-touch, sensitive to pressures and flow, multi-user, and much more intuitive—hence, a Natural User Interface.

There is a fun and informative online community that is constantly growing as more interest picks up in multi-touch research. To see some of the great sites and technologies we’ve come across, visit our Resources page.