Media Interaction Lab

Like my previous post on my multi-touch experiences at this years Cebit trade fair I also want to sum up my recent experiences with multi-touch projects at the Laval Virtual Exhibition. While virtual reality still being the main focus of the exhibition, interfaces always have been a very important part of this come-together of the virtual reality community. At Laval Virtual 2007 our project “Shared Design Space” was pretty much the only large scale interactive table that enabled accurate multiuser interaction (video). This year however multi touch is big and it is also big in Laval!

Warning: Detailed information ahead!

More than 8 interactive table setups could be seen at Laval Virtual based on various technologies like the MERL Diamond Touch, the ever popular reacTIVision tracking framework and of course FTIR (and DI). In this (not so) short summary I want to focus on optical systems since they rely on both custom built hard- and software, namely 6: Unfortunately there is no high res fotos this time since I only took videos… which I will eventually cut together to a short clip some time. For the time beeing I give you screenshots of the videos!

Interactive Scape: (at the MOBIDIUM booth) (commercial, FTIR)

Here it was again, the white beauty from Werk 5. And it’s not only beautifully made, its also bloody heavy :). I happened to have a chance to help putting it back in the transport crate after the exhibition. Well, it took three strong men and me to lift it back to its secure box. The table now also integrates communication with Bluetooth enabled devices like mobiles, which unfortunately wasn’t shown. The demo application, which is developed in VVVV looks like its still the one from CeBit and only allows basic interaction with displayed images.

Some more news on the projection quality which I criticized for its low brightness at Cebit: It was only running at half brightness then to achieve longer lamp-life. Seems strange since I also got some information on the price (which I won’t give away) and that should be enough change to get some spares! Still, best packaging seen so far (except for our FLUX-table of course ) and still the best resolution also… great work!

Table at MOBIDIUM booth (commercial, DI)

MOBIDIUM also showed a second table which was using the reacTIVision tracking framework which was originally developed for the reacTable (video) project. The table was actually finished the two days right before the exhibition started, resulting in building noise during setup :), some sloppy hardware design (e.g. holes in the table to fit in the projecotrs) and only mediocre software (some music/particle application). Still the hardware setup is interesting, featuring two Toshiba short-throw projectors (Model ew25) and a total resolution of 2560*800. Also the tracking of the tangible interface-tiles worked pretty well, the showcased software however was not that impressive.

Intuilab (commercial, FTIR)

Intuilab too presented their multi touch table (video) at Laval Virtual. The guys were showing the most advanced image viewer application of the whole exhibition with some pretty neat options like intuitive tagging, timeline and presentation mode functionalities and integrated google-earth viewer. Tracking is based on an extended version of the Touchlib and the application (actually the whole framework) was written using Perl (!). Pretty fancy stuff :). Tracking was good, the application however seemed to be a little slow at times, probably due to the interesting software development language. Last but not least: “Rotate and Translate” rocks!

Immersion (commercial, DI)

The only real multi touch table at Laval Virtual using DI, enabling also some additional functionality like detection of real obstacles on the screen. This way physical objects like mugs or business cards can be placed on the table resulting in all the images projected underneath moving away. In addition to that images could be “thrown” to a second monitor which worked as some sort of presentation surface. The whole application however was lacking finesse and very much looked like some proof of concept demo. Given the fact that the touch tracking worked pretty well and that it can easily be extended to feature marker tracking the hardware setup looked quite promising, the software however needs improvement. (like most really!)

byVolta / NUI (commercial, FTIR)

Finally one of the famous NUI-tables under my fingertips, presented at the booth of byVolta. NUI (famous also for their great nuigroup forums) showed a whole set of applications embedded in a framework. A special gesture could be used to bring up a menu, enabling users to switch between applications and more. Only the google earth application was shown separately. The NUI table uses a advanced version of the Touchlib (called touchCore) to track whats happening on the table, with pretty good results. Also the window-system is impressive with its possibilities to run several applications at a time. This is state of the art multitouch - software with great concepts and only little flaws. The table itself looks like most of the other tables with uninspired, boxy design. (Well, thats not entirely true… it has the nui-logo on each side which looks great!)

And last but not least and my personal heroes:

ESIEA (student, FTIR)

This project (video) was the only multitouch project which was only research and was developed by several students. Why I think this project is great has several reasons:

-great new software (games to be exact), no mere “jeff-han-clones”
-inspiring easy, robust and low-cost hardware setup
-the student’s skill of playing the games the created (I totally sucked)

The hardware setup was probably the boxiest one I have seen so far with an projector used that can be best described as an extreme narrow angle projector. The projector was actually mounted above the table-surface, projecting through a hole in the table onto a mirror. The mirror, which was just lying on the floor of the table with no mounting whatsoever (not even tape) reflected the image back up on the rear projection screen. This is especially interesting since as second project of the same university which was displayed right next to the table used a short throw projector (Toshiba ew25 again).

This might make you think where I liked the hardware setup in the first place: Well, mounting the projector outside the box gives you interesting opportunities. When using a tilt-able mounting you could also project onto a nearby wall, e.g. for presentations when the table is not needed. This way the flexibility of the whole system is increased and therfore actually make a virtue out of a necessity. This is especially true when thinking of current HD-projectors throw ratios which aren’t suitable for rear-projection systems. Also they only used 30 LEDs for a pretty big table and still claimed that it was still too much! And last but not least the went for a really cheap web cam and custom tracking software with excellent results!

Also they did not use any compliant surface whatsoever (In case you tried it you might not even have noticed). They just avoided the whole quest for a finding a good compliant surface by changing the interface metaphors to their needs. The game simply didn’t require any dragging and still was great fun!

This brings me to the next point right away. The games they did were really enjoyable! (although I was really bad at all of them) Inspired by some games like “dance dance revolution” and “guitarhero” which were adapted for multi touch, the games were both appreciated by children and professionals alike. This is the way multi touch should be done: strictly underground, strictly unbiased and creative. Thanks again for the enjoyable games and hope to see more soon!

Well, this concludes my summary of tables in Laval. Thanks for staying with me if you have read the whole thing, and if not… well, you are probably missing out something!

Oh, one more thing that was interesting: an acoustic multi touch table!

There always seem to be tracking technologies no one has ever thought of before and I believe this is one of them: For the project “Artefact” the developers used a special acoustic tracking which detects if someone is knocking on the surface with his/her fingers. The surface itself is made out of some sort of plastic and has a artistically curved surface. Under the surface two microphones are attached which record the sound of the finger knocking on the surface. Depending on the surface thickness on that specific point and on the position on the surface the microphones pick up the signal with different intensity and delay. Comparing this information to a prerecorded set of data the system can determine the spot on the surface.

Where this data comes from you ask? Well, each point has to be configured separately. This means that if you want to calibrate an interactive surface of 20 * 20 cm with 1 cm resolution you will have to knock on the plane 400 times on all wanted positions and record each single time to aquire all necessary data. According to the developers it is also necessary to have roughly the same temperature when you want to use the system as you had when calibrating it :).

By the way, the system exhibited was much larger than 20 by 20 centimeters (with an resolution of 2 cm though) which must have been a real pain to calibrate! Still, it works surprisingly well and could also be adapted to be more accurate and even react differently to various objects you use for tapping on the screen. (You would have to calibrate each object separately though). Creative work!