First of all i want to say this is a great community, learned a lot from it.

I came across this yesterday and am trying to understand how it is done.

here is a video of it:
http://www.youtube.com/watch?v=C5pGyB2L71o

a small summary of their technology:
http://www.flatfrog.com/technology

I also found this, not sure if it is related.
http://www.planar.com/advantages/whitepapers/docs/Planar-AMLCD-Optical-Touchscreen.pdf

Does any one know anything about this technology and how it works?
Any help is greatly appreciated.
thanks

here you can find a clear answer :

http://www.touchscreenmagazine.nl/multitouch-techniques/planar-scatter-detection

the PSD is just an adaptation of ftir

http://nuigroup.com/forums/viewthread/1731/ try hear

BenjaminV - 11 February 2010 05:01 PM

here you can find a clear answer :

http://www.touchscreenmagazine.nl/multitouch-techniques/planar-scatter-detection

the PSD is just an adaptation of ftir

Thanks that is nice information, But they claim their product has “Ambient light immunity” and “can also handle object size recognition”. And i don’t understand how they are able to detect the reflection on large scale in all kinds of light without cameras.

Edit: Also they claim that pressure can be detected by their product as well.

gogo - 11 February 2010 05:05 PM

http://nuigroup.com/forums/viewthread/1731/ try hear

I am sorry but i don’t see how that relates.

First post from me.

Coll, I believe that Touchscreenmagazine.nl is incorrect in their understanding and explanation of the technology.

You can find some of their patents online - their technology being developed from two prior companies that folded: Open and Taktio and now Flatfrog.

Anyway, the light is being emitted from a backlight using IR (I saw one online posting about using 980 nm wavelength light) through the LCD and a front surface (waveguide). The principle is not technically FTIR since no light is totally internally reflected in the waveguide absent a touch on the waveguide surface that causes the light to scatter into it. Then cameras positioned at the corners of the waveguide each with an pinhole aperture in front is able to detect position based on where the light scattered down from the touch point is hitting the CCD and it’s all geometry from there to find intersections. With 20+ touches, they may be using some additional techniques, but at least need some great algorithms.

It’s actually not a new technique, there are a number of patents and implementations of this sort going back 20 years but a very good implementation of it.

Check out US patent #7465914 for more about it. That’s my $0.02.

Maybe I will do some more research and write some more about it on my new blog: http://www.multitouchinsider.com

Hope that explains

To the other posters: I think Coll is wondering how they can make a screen that flat.

PSD explains how the light gets to your fingers, but not how it gets from your fingers into the computer with that thin of a screen.

If it is truly as flat as it looks, these screens can’t work by a single camera. They probably have a multi sensor setup with 1 camera for each small subdivision of screen. This would enable the touch resolution they are getting, as well as making it so you didn’t have to put the camera that far back. That pdf link in the OP seems like it is an integration of an image sensor and TFT screen in one unit. I don’t think they are making 50” diagonal optical sensors though. They might be, who knows?

JF

So dean, if I understand correctly it is kind of like reverse FTIR. The light source is perpendicular to the screen plane and the camera sees through the edges? Brilliant, but sounds hard to implement.

Essentially the camera gets a blob that only moves on the X axis and by triangulation (quadangulation?) the position of each blob is determined.

Side note: I love how the flatfrog website measures latency in Hz, funny marketing people.

Thanks for all the information!

deanEkko: That is interesting. The patent also had a lot of useful information. Thanks! I also bookmarked your blog

jonferran: Thank you as well!

So basically from my understanding it is something like the quick sketch-up i made using paint. (Attached) With 2+ cameras

Only thing i don’t understand still is how is it that it is unaffected by other light, such as direct sunlight. Maybe all the cameras have bandpass filters? But still it wouldn’t be as affective as they claim. Also I don’t see how they can get pressure using this technique. (As they claim)

jonferran, yeah, reverse FTIR would be one way to describe it.

This could be a great way for the NUI community to reduce the sizes of the hardware setups. You could effectively make a camera-based FTIR system that normally needs the depth for the field of veiw for the camera, down to the thickness of an LCD. It’s not really any more difficult of a hardware setup, just not as obvious of a way to approach it.

I think the answer to coll’s question might be this:

The reason camera-based FTIR systems are so deep is because of the field of view for the camera. If you’re making a system with the LED light injected into the edge of the acrylic, there isn’t much offset for the light source, but a lot for the camera (and projector ?).

Turn the whole setup on it’s head.

Place cameras against the edge of the acrylic and use a rear-DI setup for the light and you have the essence of FlatFrog hardware.

That’s bound to be thinner than a regular FTIR setup.

I think it is immune to outside light because outside light doesn’t have anything to totally internally reflect off of when it hits the top surface. It simply passes through the screen and doesn’t get picked up by the cameras.

Here’s a crappy paint diagram of how I understand this working.

jonferran, exactly right.

Take a look at Snell’s Law as applied to Total Internal Reflection (TIR) and you’ll see that it is impossible for ambient light to enter the waveguide and become internally reflected from an angle above or below the flat plane unless it has something to scatter off and become trapped.

Some ambient light will reflect off the front surface (from the sun in your above drawing) and some will enter the acrylic and reflect back out from the second surface but no light will get trapped and TIR inside.

Maybe a technique like this would solve some common problems that people in the NUI community have with ambient light, bandpass filters, etc.

You could probably also pull off such a technique using the regular LCD backlight but then what about when the screen is dark vs light, you would get different signals… something to research.

Thanks for bookmarking my blog jonferran - still working on the site, any feedback welcome

oh that makes sense, thanks! Sounds like it would be a nice project to do.

coll,

right on, using pinhole apertures, only direct lines from the touch point to the sensor will make it and all the other lines will not be able to make it through the aperture.

From what I understand, this technique was really a precursor to FTIR and had been a lot in the past. I’d be interested to know what limitations are considering that FTIR was/is the predecessor.

I would be curious to see what the camera sees since it has no direct view of the object. It must see a weird reflection coming from the inside surface of the acrylic.

is it something like this?

JF