DTphonehome said:
It's actually pretty clever. A typical multitouch display is very complex...what this one does is sense the disruption in internal reflectivity of the uppermost layer of the screen. The "camera" is not exactly the sort which you'd use as a video or still camera...it's more like the "camera" in your optical mouse.
In the case of the cited demo video, it is.
See
this page for an explanation of the tech. Note the video clip in the upper-right corner that shows the raw camera output overlayed on the device.
Their FTIR system is a variation on how fingerprint scanners work. A light source shines edge-on into a glass panel. An ordinary (B&W) camera facing the panel picks up the internal reflections from this light source interacting with touches.
DTphonehome said:
It senses when a new point is touched and detects the track of that point by the light being scattered as it moves.
All done through software examining the image. Looking at the video, it appears that touches show up as very bright spots, while surface dirt is fairly dim, making it fairly easy for software to detect touches.
The tracking is a bit more complicated, but can also be done in software if the computer is fast enough.
DTphonehome said:
The limitations of a classic touchpanel is that it is essentially a grid. When you depress a portion of the grid, the computer sees that as a point based on the x and y coordinates.
This isn't the problem. The problem is that classic touch screens do not have a grid. They use two charged surfaces and a very small number of sensors (four or five) to measure how resistance or capacitance between the surfaces changes at the sensors. This raw data is processed to produce coordinates. Which is why multiple touches don't work.
If there was actually a grid of wires, then you could detect multiple touches, in much the same way keyboards (which often use a grid with keys at all the intersections) can detect multiple simultaneous keystrokes.
