Build your own Surface table

[RayB]

I'm considering building one of these. All the software is open source or free (except Windows of course). You can code your apps in your preferred environment, even Flash. I would make a small coffee table that acts as a media center, remote control for devices, etc. A cool addition to an entertainment room.

http://www.maximumpc.com/article/features/maximum_pc_builds_a_multitouch_surface_computer

[SavannahLion]

You have got to be kidding. That seems almost too damn easy.

What I don't understand though is why the PS3 camera in particular used for this? What sets it apart from other cameras? Hackability?

[MonMotha]

Somebody at (former) work played around with this, though they ran it on Linux, not Windows.  The choice of camera comes down to finding one that can be made to be sensitive to IR.  Many camera sensors are sensitive to IR, but they have filters over them to filter out the IR since humans are more interested in visible light when it comes to webcams.  Removal of this visible spectrum filter and replacement with an IR pass filter stolen from a TV remote receiver or similar does nicely.

Biggest problem I noticed was slow "recovery time".  That is, if you pressed on the surface and let go, it took a while for everything to return to normal and the press to fade out.  That's not too bad for most normal desktop use (though it limits maximum "double click" speed), but for some applications it can prove troublesome.

Of course, you also have to do rear projection in order to get enough room for the camera.  Not suitable for "panel" style touchscreens.

[orion]

That looks slick and surprisingly simple. If you build one, please document it, that would be a cool project. I like the idea of using one as a remote... that would be cool, and well the ultimate in uber geeky.

[massive88]

What about incorporating this on a smaller scale into a CP build?

For instance, joysticks as normal, but where the button layout goes, would be a plexi screen with LEDs to light up "buttons" based on your game selection.

Any lag in registering "presses" would be a downer, but for a modular layout, that would be totally sweet.  It would be like a touch screen modular button layout, and just using LEDs to illuminate the buttons as opposed to a projector would make it a fair amount cheaper.

[SavannahLion]

It's a cool idea, but you would lack any tactile and positional feedback. As soon as your hand floats away from the virtual button, you would be pounding on the plexi in frustration.

Given that the size of the plexi is limited only by the projector and distance, perhaps one could create a dynamic monitor, touchscreen and bezel?

Let's say you want a 19" 4:3 monitor. That gives you ~15.2"x~11.4". If we make those numbers nice and round, we'll make our bezel that expands the total size to 25" x 25" giving us a the equivalent of a ~35" 1:1 monitor. This gives us ~5" of bezel on the left/right and ~6 3/4" on the top/bottom. That gives one enough room to dynamically assign instruction cards and probably enough room to display just about any bezel (additional research required to determine if a 5" side is enough space) one heart desires. Added bonuses would include an auto rotating monitor, expanding the 19" play area to the full size of the viewing area at the cost of bezel space, inclusion of any touch screen game (baring lag issues of course), a juke, etc. etc.

The only catch is working out the necessary resolution one would need to effectively make such a display.

[Ummon]

Well, as far as that goes, once your hand moves away, you're not pounding on an physical button, either. I thought of this idea using small LCD panels, a few years ago. And on that note, isn't the Win model just a glorified LCD? Don't they make touch screen LCD panels, or at least LCD units? Of course they do. Jukebox at the bar. So make a bigger one. I don't see the thousands of dollars in this that these things are sold for. Pshaw.

[massive88]

Multi (infinite) touch as opposed to single touch I think is the difference.

[SavannahLion]

Well, as far as that goes, once your hand moves away, you're not pounding on an physical button, either.

That doesn't make any sense. You somehow inferred the problem I presented as being a valid problem with current button use. You can feel where buttons are, you can feel their movement, bounce, and (with some) switch engagement. You can touch a button without triggering it. You have to touch a touch screen to trigger it, which would prompt you to have a lightest of feather touches not to trigger. Since not everyone is going to have such a light touch, it would serve to add to the frustration for people who are "heavy handed".

With nothing to feel on a touch screen, chances are really good your fingers are going to float from button center. Not so bad if the interface is at least consistent like the Virtual Laser Keyboard, but when you have a dynamically changing interface that something like this is going to inherently have, people are going to trip up.

Think about the principles of how any touch screen operates. You have virtual buttons presented with a visual representation. Point is people are looking at the "buttons" they're pressing. Why do you think Nokia (or whomever they contracted out) spent so much time and money on a tactile touch screen?

[Ummon]

Multi (infinite) touch as opposed to single touch I think is the difference.

I thought it was a matter of software.

That doesn't make any sense. You somehow inferred the problem I presented as being a valid problem with current button use. You can feel where buttons are, you can feel their movement, bounce, and (with some) switch engagement.

Ah, right. Thanks for correcting me on that.

[MonMotha]

Multi (infinite) touch as opposed to single touch I think is the difference.

I thought it was a matter of software.

NO.  There are fundamental hardware differences between most multi-touch and most single-touch designs.  The reason mostly boils down to cost: the methods that allow only a single touch point are cheaper than those that allow multiple.

Most common touchscreens which do not require a special stylus are resistive.  They basically electrically look kinda like a potentiometer where the "wiper" position is determined by a pressure point.  Of course, there are two axes that look kinda like this so that both X and Y position can be sensed.  It's electrically impossible to distinguish multiple individual pressure points on this.  It just looks like the average of all of them weighted roughly by how hard each point is being pressed.  Resistive touchscreens are popular because they're relatively cheap and easy to talk to.

Resistive touchscreens can be made fairly large.  20 something inches is a good practical limit.  30-40" would certainly be possible, but there's not much demand for it.

Most common "multitouch" (and note that words is a trademark of Apple, or at least it is with certain casing) touchscreens on mobile devices are capacitive.  This costs quite a bit more to produce as you need to create a uniform electric field across a clear surface.  This requires a somewhat conductive clear material.  These materials exist, but they're not cheap.  You also need a pretty accurate method of measuring that field - like 18 bits accurate.  The relationship between measured value and position isn't linear, so it takes a surprising number of bits to achieve desired accuracy over the entire surface.  All this adds up to $$$.  They also can't be actuated with anything other than a grounded, conductive object, or a body part (which looks kinda like ground due to its size and electrical presence).

It's also possible to make a capacitive surface that can only sense a single touch point.  It's a little cheaper, but not a whole lot.  The military has been known to use them because they're pretty rugged, but they may be moving to strain gauge based systems as commonly used on things like ATMs, now.  I would guess that many jukeboxes also use this system as it's pretty darned rugged (like "hit it with a hammer and it doesn't really care" rugged, if you want).  Strain gauge systems cannot be made multi-touch without greatly sacrificing ruggedness, and even then it would be rather limited.

Making a conventional capacitive touch surface the size of a table is not presently practical at any cost.  There are electrical issues that aren't really solvable using present technology.  I have actually looked into it.

There are other methods of making a screen surface capable of sensing multiple pressure points.  The "Microsoft Surface", which uses a technique known as "frustrated total internal reflection", is one of them.  Downsides have already been mentioned.  The major upside is that it's not terribly costly, but costs do still add up for large areas if you want it done "well".  There's a big tradeoff in materials between image quality and how well the touch surface works.  Getting it "good enough" on both fronts can cost.

[massive88]

Multi (infinite) touch as opposed to single touch I think is the difference.

I thought it was a matter of software.

Heh no.  Thats why the iPhone was/is such a big deal.

[Ummon]

NO.  There are fundamental hardware differences between most multi-touch and most single-touch designs.

There's a big tradeoff in materials between image quality and how well the touch surface works.  Getting it "good enough" on both fronts can cost.

Oh. Well thanks for the extensive explanation. I guess the design posted here is pretty good then. Looked pretty good to me.