21. What if you replaced the tip with a ir led + sensitive switch? It could then be a real pen! So many possibilities, so little time...

Posted at 5:26PM on Dec 10th 2007 by Harrison

22. What a great application! Nice demo of its capability also.

Posted at 7:04PM on Dec 10th 2007 by Alan

23. You can use a regular red led, not just IR. The wiimote is sensitive of primarily IR but red led's contain a high enough content of IR to work in my experience. Just get a clear red LED (one where the emitted light is red as opposed to just a frosted or "painted" red lens)

Posted at 7:38PM on Dec 10th 2007 by Jordan

24. I do find this terribly amazing and no doubt I'll be trying it out this weekend, but a question I must ask is this:

Why only now are these kind of hacks coming out? Could this not also be done with a webcam with some fully exposed negatives over the lens? Maybe the Wiimote made it more practical by giving us the relatively cheap battery operated IR cam with bluetooth built right in.

Posted at 9:52PM on Dec 10th 2007 by Mike

25. This project isn't substantially different from the one that I did that was featured here recently:

http://www.hackaday.com/2007/11/02/pysight-post-halloween-entertainment/

I just used an iSight instead of a wiimote.

Posted at 11:03PM on Dec 10th 2007 by a random John

26. That's incredibly cool. Also, I have to commend the guy on his presentation skills. Much better than these kinds of things normally are.

Posted at 12:19AM on Dec 11th 2007 by tiuk

27. To those asking why the wii mote and not a webcam is distributed processing. The WIIMOTE is what handles the ir to space computations, not the computer. A webcam does not do that. It's like video cards with hardware mpeg2/4 decoders. The wiimote allows for up to 4 different ir sources being tracked and turned into x,y calculations. The whiteboard and multitouch just map those x,y coordinates and move the mouse based on that.

Posted at 12:23AM on Dec 11th 2007 by cde

28. #20: if you do that make sure you get the voltage right.. 0_o

Posted at 10:37AM on Dec 11th 2007 by Tim

29. I had done visual tracking with a web camera and opencv.
http://youtube.com/bluehash

But the wiimote way is a much sturdier way IMHO.

Posted at 5:04PM on Dec 11th 2007 by Gerard

30. what about an IR laser... would mean that you could control this from the other side of the room. i'm seeing this in my living room sometime soon

Posted at 5:49PM on Dec 11th 2007 by slaintemiq

31. Linux support please? :)

Posted at 9:46PM on Dec 11th 2007 by Furthur

32. Mary, mother of God...

All hail the Wiimote!

Posted at 10:24PM on Dec 11th 2007 by Mr Black

33. I love this thing! I put one together today, and it works perfectly! Thanks Johnny, if you're reading this!

Posted at 10:38PM on Dec 11th 2007 by Aklyatne

34. random john:
On your blog you mention the camera calibration step...
"The game blanks the screen and then shows a red dot in each corner. The camera captures the location of these dots and then uses that information to translate the laser location from camera coordinates to game coordinates"

Any chance you could do a write up on that? Maybe release your code? I've been trying to work out how to build the projection matrix but all the info on the net is way above my head.

Posted at 10:46PM on Dec 11th 2007 by dam

35. Man! That was sweet! I made two pens last night and it works great! I used it today for presentation in my class Human Computer interface design.

Posted at 11:37PM on Dec 11th 2007 by David

36. So is there a tutorial on how to build a pen? I sure want to try this...

Posted at 12:25AM on Dec 12th 2007 by ken

37. WOW Simply Amazing! This dude is pretty crafty! Nintendo should hire this guy to make the next Wii! ;-)

Posted at 12:27PM on Dec 12th 2007 by R.MnTnA

38. dan (#34),

I'm going to be presenting on the thing at PyCon and opening up the source at that point. For now I'll give you the quick and dirty explanation.

I worked out the math for doing an accurate four point calibration but before I implemented that I decided to implement a very naive version that wasn't nearly as complex. As it turned out the simple version was accurate enough for my purposes, especially given the fact that the camera is well aligned with the projector so I don't get much keystoning with either device.

What I do is put a red dot in a corner until the camera "sees" it and then proceed to the next corner.

Then I define a bounding box within the camera view. I do this very simply by taking the topmost Y value as the top of the box (it could have come from either the upper left or upper right) the leftmost x value as the left of my box and so on. So within the camera's 640x480 view I have a smaller area that is mapped to the screen. It might be from (80,100) to (500,389) or something like that.

If my game surface is 800,600 then I need to translate a point from that box to the 800,600 surface. Let's say that I find the red dot at (400,220) in the camera's coordinate system. First I subtract (80,100) to get (320,220). Then I divide by the width of the box (480,289) and multiply by (800,600) to finish the translation. In this case the result is (533,456).

If I had a lot of keystoning then I'd need to do use a more accurate transform but the quick and dirty one was always within a few pixels.

Posted at 1:52PM on Dec 12th 2007 by a random John

39. if you remove the IR protection off of a conventional laser pointer get two of them you could do the same thing right? but from far away.. and not have to stand up.. you could even attach them to your head and have a remote to turn them on so you could just point you head where you want to go and mano.. im gona try it..

Posted at 5:48PM on Dec 12th 2007 by ioniser

40. I hope many are trying out this brilliant thing, because so am i.
I also hope someone (with the knowledge about) could set up a tutorial about how-to make a fine working pen!
looking forwards, Michael.

Posted at 6:30PM on Dec 12th 2007 by M. Scofield