My WiiMote project is complete!

[CyBerkut]

Johnny did a great job of publicizing wiimote homebrew by making some great videos and apps but his work was not entirely new and was largely founded on the hard work of others. I believe the guys that reverse engineered the wiimote and made the details public, making all this possible, deserve much more credit.

 

Positioning the camera is probably the biggest pain for this project, it has to be rock solid for calibration reasons but is likely to get in the way and be knocked.

 

Ah, good point about Johnny's main role. I stand corrected! (Which happens waaayyy too often... :book:) Kudos to JLC for promotion, and the developers (whose names I do not currently know) for their work.

 

Solid camera positioning is indeed important. As I recall, in a JLC video, he had it more off to the side, rather than behind / over the shoulder. Of course, I've been asleep since then so I may have that wrong.

 

The more I look at this (and headtracking which I have working), the more appealing it becomes to take down the ceiling fan and go with ceiling mounts for the Wiimotes. :smilewink: FreeTrack would need to have some axis rearranged, but it would be a view with less opportunity for obstruction. And it would be relatively simple to rig up DC power supplies to the Wiimotes so that batteries would no longer need to be changed / recharged.

[Joe Kurr]

Hehe, I'm working on exactly the same solution here!

When I first saw Johnny's videos I bought a Wiimote and some IR LEDs and started experimenting.

I haven't had time yet to build a glove, I'm stuck to holding the LED in front of the screen and connect a wire to a battery in order to do a mouse-click.

 

Another nice aspect of this system is that you can turn anything into a clickable panel, just point the Wiimote to the surface you want clickable, calibrate, and you're off!

I'm thinking about trying this with my cockpit :)

 

When I get results, I will post them here.

[Winder]

No. The wiimote camera is very powerful capable of tracking all four points independently at 100 FPS. You can increase the cameras FOV by placing it further away and using high output Infared L.E.D.'s 1000+nm should be ample for 2-5 meters.

Moving further away doesn't change the camera's field of view. That is always static, based on the lens. You're correct that you will have a larger area to operate in, though. But, then you're sacrificing a great deal of fidelity if you have a lower resolution camera. The Wii remote has a resolution of 128x96 (http://en.wikipedia.org/wiki/Freetrack), which is painfully low. At that resolution, the pixels will represent huge blocks of physical/3D space (like maybe 6 inches or a foot?)... which means that any pixels that get dropped from remotely quick movement will result in huge jitter.

 

anything over 35 fps the human eye will see as fluid movement.

Mmmm.... this is only partially true, and I think not relevant to marker tracking. It's true that the human eye can't tell the difference between 35 FPS, 50 FPS or 500 FPS when it is viewing a video, but this kind of "FPS" is totally different. You're talking about how sensitive the eye is to an updating image. This FPS is related to how fast the camera is refreshing to track movement (and, accordingly, how in line it is with the video card/game). Higher FPS is always good, essential, and very noticeable in point tracking. Even the FreeTrack software accounts for the crappy frame rate in many webcams with its interpolation equation. Why? Because 35 FPS will have large gaps in point tracking with even remotely quick motion. Motion blur = inaccurate and jittery tracking. You've even somewhat addressed this in your statement below (saying that the Wiimote is superior to webcams because of its higher FPS).

 

*EDITED* Don't use Webcams with FT 2.2 they are inferiour to the IR camera in the wiimote (30 FPS v's 100 FPS) and a waste of time IMO.

Another way to look at it… imagine you’re watching video of a golfer practicing his swing. Let’s say his club head is being tracked by markers using a motion capture system. You’re watching the results on a monitor that’s refreshing at 35 FPS (5 FPS faster than television, I believe). First, his club head is tracked by cameras that update at 15 FPS. Then his club head is tracked by cameras that update at 400 FPS. In both instances, the monitor you’re watching on is refreshing at the same rate. But the tracking quality is going to be vastly different. Why is this detectable to the human eye? Because the FPS is not impacting what you see from your monitor. The FPS is impacting what the monitor receives. This concept is exactly the same as what happens with head tracking. Cameras w/ low FPS don’t accurately detect motion real time, which means that by the time your see the in-game view with your eyes, what you see is already outdated.

 

I also use a Wiimote with FreeTrack 2.2. The FreeTrack folks have documented that the Wiimote IR camera puts a much lower load on the processor than a web cam does.

Probably because the Wiimote's resolution is atrocious, which means that most processers probably make light work of it. Maybe, additionally, the Wiimote does some processing on its own board? Not sure on this.

 

I won't opine that FreeTrack with a Wiimote is just as good as a TrackIR because I don't have a TrackIR to compare to. I will venture that based upon a comparison of the TrackIR's camera specs vs the Wiimote's, that a properly adjusted FreeTrack 2.2 using a Wiimote should be entirely acceptable.

It sounds like many FT users think that the Wiimote is the way to go. I'm going to try it myself (hopefully today or tomorrow, but maybe not for a week or so with holiday plans fast approaching), but I'm skeptical that it will be better than a good Logitech webcam (since its resolution is so much better). Certainly a TrackIR has way better specs, all around:

 

Wii Remote

Resolution: 128x96

FPS: 100

FOV: 41 degrees

On board processing: yes (I think)

 

TrackIR 4

Resolution: 355 x 288 (but, with their subpixel horizontal processing, it actually is accurate to 710x288)

FPS: 120

FOV: 46 degrees

On board processing: yes

 

*Specs taken from http://en.wikipedia.org/wiki/Freetrack and www.trackir.com

That Wiimote resolution is really troubling, especially if you take into consideration the resolution per degree. That would be about 3.12 pixels/degree with a Wiimote compared to about 4.06 pixels/degree with a TrackIR 4. If you're moving more than two feet or so away from your camera, 3.12 pixels are going to be handling very large amounts of real estate... even if you use high powered LEDs. It really seems to me like JonK's project might be even more accurate with a higher res camera--be it a webcam or TrackIR or whatever else is on the market.

 

Prior to seeing JonK's thread here, I was thinking that I would locate the activation switch differently. My first thought was to put a switch on the middle finger tip pad, such that it could be activated by the thumb while pointing the index finger (with the IR LED mounted upon it) at the screen. That should allow the IR LED to be right on the tip of the finger, hopefully yielding good, repeatable accuracy.

I like this idea. It would make me less nervous about damaging my LCD screen. I'd probably miss the touch feedback, though.

 

Regarding the use of Wiimote(s) with a PC, be aware that Bluetooth stacks can be a *major* issue. My Acer Aspire has Bluetooth built in. While the included Bluetooth stack appeared to yield a "connection", it did not achieve useable communications with the Wiimote. (Vista 64 bit, BTW). The only stack that worked in my case was BlueSoliel... so there was a bit of extra expense that I hadn't anticipated. If you're buying a Bluetooth dongle to add Bluetooth to your computer, I would definitely recommend looking for one that includes the BlueSoliel stack.

Thanks for the tips on this. I'm trying to get a Wiimote up and running on my PC for some experimentation. This will help. :)

 

As for using a single Wiimote to handle both FreeTrack and touch screen emulation, I don't see how the FreeTrack software would be compatible with that. In my opinion, you would need to write new software, and it would not be a trivial undertaking. By the time you got done with it, the expense of a second Wiimote would look miniscule (if you value your time at all).

Agreed. This is what I was trying to convey when I was talking about FT/Wiimote not likely being able to track multiple rigid bodies. I think that it *could* be done, but a whole new program would need to be created.

[therion_prime]

I managed to squeeze a switch (from a dead X52), a IR-LED and a "button battery" into the cutoff fingertip of an old leather glove.

 

Thanks to JonK for making me aware of this technology. :thumbup:

(I was about to order a 17" touchscreen - you saved me a LOT of money!)

 

It works great and I have to say that it's the most important addition to my flightsim hardware since I bought my firstTrackIR a few years ago!

[Vault]

Moving further away doesn't change the camera's field of view. That is always static, based on the lens. You're correct that you will have a larger area to operate in, though. But, then you're sacrificing a great deal of fidelity if you have a lower resolution camera. The Wii remote has a resolution of 128x96 (http://en.wikipedia.org/wiki/Freetrack), which is painfully low. At that resolution, the pixels will represent huge blocks of physical/3D space (like maybe 6 inches or a foot?)... which means that any pixels that get dropped from remotely quick movement will result in huge jitter..

Intresting discussion. Agreed. FOV does increase cubic capacity in relation to distance. Wiki has published false information in regards to the Wiimote it has a resolution of 1024x768. The FT dev also states this on the FT website. (even Wiki advise's on that link that this resolution has been reported) With that type of resolution distance's of upto 5 meters will be visible and very accurate tracking on the Wiimote's camera.

Mmmm.... this is only partially true' date=' and I think not relevant to marker tracking. It's true that the human eye can't tell the difference between 35 FPS, 50 FPS or 500 FPS when it is viewing a video, but this kind of "FPS" is totally different. You're talking about how sensitive the [b']eye [/b]is to an updating image. This FPS is related to how fast the camera is refreshing to track movement (and, accordingly, how in line it is with the video card/game). Higher FPS is always good, essential, and very noticeable in point tracking. Even the FreeTrack software accounts for the crappy frame rate in many webcams with its interpolation equation. Why? Because 35 FPS will have large gaps in point tracking with even remotely quick motion. Motion blur = inaccurate and jittery tracking. You've even somewhat addressed this in your statement below (saying that the Wiimote is superior to webcams because of its higher FPS)...

The adult human eye cant see no more than 35 FPS on a screen. The Wiimote and FT work in real time so if you get 100 FPS from the Wiimote IR camera you'll get 100 FPS of tracking from the software. Interpolation is software based and it is no way a substitute for real hardware based FPS. The Wiimote is superiour to any webcam. Even FT's Dev states it's far superiour.

Probably because the Wiimote's resolution is atrocious' date=' which means that most processers probably make light work of it. Maybe, additionally, the Wiimote does some processing on its own board? Not sure on this.).[/quote']

Atrocious? On a hardware level, camera v's camera, the Wiimote's camera wins hands down infact it's nearly 2.5 time superiour regarding resolution!, power consumption, cpu cycles also go to the Wiimote as does running temprature. The Wiimote process's all raw axsis data sending it to the FT/TIR interface via bluetooth. FOV is only limited by distance to the tracking points.

 

TIR4 runs a lower resolution camera and use's a processor and software to crank up the resolution using subpixel horizontal processing, which again is no substitue for real hardware based resolution this is why TIR feel's hot when you touch it. It's also why TIR requires nearly double the voltage of the Wiimote.

 

Now let's talk about price comparison. :smilewink:

 

TIR4: $129.95 v's the Wiimote: $30.00 - $40.00.

It sounds like many FT users think that the Wiimote is the way to go. I'm going to try it myself (hopefully today or tomorrow' date=' but maybe not for a week or so with holiday plans fast approaching), but I'm skeptical that it will be better than a good Logitech webcam (since its resolution is so much better). Certainly a TrackIR has way better specs, all around:.[/quote']

 

FT 2.2 is brilliant with a Wiimote. FT has evolved.

Wii Remote

Resolution: 1024 x 768

FPS: 100

FOV: 41 degrees

On board processing: yes

 

TrackIR 4

Resolution: 355 x 288

FPS: 120

FOV: 46 degrees

On board processing: yes

 

I corrected the hardware specs of the two IR camera's.

That Wiimote resolution is really troubling' date=' especially if you take into consideration the resolution per degree. That would be about 3.12 pixels/degree with a Wiimote compared to about 4.06 pixels/degree with a TrackIR 4. If you're moving more than two feet or so away from your camera, 3.12 pixels are going to be handling very large amounts of real estate... even if you use high powered LEDs. It really seems to me like JonK's project might be even more accurate with a higher res camera--be it a webcam or TrackIR or whatever else is on the market.[/quote']

See above hardware spec.

Agreed. This is what I was trying to convey when I was talking about FT/Wiimote not likely being able to track multiple rigid bodies. I think that it *could* be done' date=' but a whole new program would need to be created.[/quote']

Agreed. Both Ft and Wiiwhiteboard are both open source now all we need is some coder clever enough to modify it. The whole concept of using one Wiimote to do both head tracking and touch screen is limited by software.

[Vault]

I was thinking that if you Guy's who are repetitively tapping the screen with the face of the contact point will it damage and ruin the clarity of your screen? I'd at least cover the face of the contact point with some type of PTFE paint or tape and use a switch with as little mechanical resistance as possible. Repetitive pressure on an LCD screen can't be a good thing long term.

Edited December 23, 2008 by Vault

[TheMoose]

Impressive video. Basically should also work with a large touch screen, then no glove required :thumbup:

[Vault]

Impressive video. Basically should also work with a large touch screen, then no glove required :thumbup:

 

Good post although touch screen's are expensive you've given me a great idea. Touchscreens use an array of touch switches which would eliminate any risk of damage to your expensive LCD screen. With a 1.5v watch battery you could run a touch switch and a IR LED in something the size of an elastic sewing thimble which would fit neatly on the end of your finger.

[Winder]

Wiki has published false information in regards to the Wiimote it has a resolution of 1024x768. The FT dev also states this on the FT website.

 

Actually, the information that I quoted on the Wiki is correct. The specs you are referring to are based on early rumors that circulated before more in-depth reverse engineering was done on the Wiimote. While the Wiimote sensor's specs are not publicly circulated, I think it's pretty common knowledge that the most up to date information has the Wiimote sporting a 128x96 resolution. (See http://wiibrew.org/wiki/Wiimote#IR_Camera for another source... the go-to source for Wii hardware info.) The 128x96 resolution is the actual sensor resolution. Just like the TrackIR 4 has an actual sensor resolution of 355 x 288. The Wiimote's 1024x768 is actually the result of 8X subpixel analysis. Much like TrackIR 4's 710 x 288 effective resolution is a result of horizontal resolution doubling through subpixel analysis.

 

The adult human eye cant see no more than 35 FPS on a screen. The Wiimote and FT work in real time so if you get 100 FPS from the Wiimote IR camera you'll get 100 FPS of tracking from the software. Interpolation is software based and it is no way a substitute for real hardware based FPS. The Wiimote is superiour to any webcam. Even FT's Dev states it's far superiour.

I agree that the 100 FPS of the Wiimote is far superior to that of a typical webcam. I was saying that a typical webcam, with vastly greater actual sensor resolution, is superior in that respect to the Wiimote. Basically, if you want acceptable resolution, you need a webcam. If you want acceptable frame rates, you need a Wiimote. If you want both, you need a TrackIR.

 

it's nearly 2.5 time superiour regarding resolution!

This isn't correct. As I noted, the actual sensor resolution is 128x96 for the Wiimote, which is practically Medieval. (Though the relative weakness of the hardware makes sense, considering what the Wii was made for... little kids and grandparents playing virtual ping pong together.

 

The Wiimote process's all raw axsis data sending it to the FT/TIR interface via bluetooth. FOV is only limited by distance to the tracking points.

According to a previous poster, the Bluetooth connection can result in some issues. I can't speak to these. FOV isn't limited only by distance. It's determined by the lens. We've already been over this though. :)

 

TIR4 runs a lower resolution camera

Again, this is incorrect. Here, I'll send you to FreeTrack's lead developer to confirm what I'm saying:

http://www.free-track.net/forum/index.php?showtopic=482&message=5033

 

and use's a processor and software to crank up the resolution using subpixel horizontal processing, which again is no substitue for real hardware based resolution this is why TIR feel's hot when you touch it. It's also why TIR requires nearly double the voltage of the Wiimote.

The Wiimote uses 8X subpixel processing. TrackIR 4 uses 2X on horizontal only. So, if there's no substitute for "the real thing" (which I don't necessarily agree with, as subpixel analysis allows for much more consistency by integrating more stable averaging math in the tracking equations), why is it better to have 8X than 2X? :)

 

Now let's talk about price comparison. :smilewink:

 

TIR4: $129.95 v's the Wiimote: $30.00 - $40.00.

Don't forget the Bluetooth dongle.

 

I think that the pricepoint sounds about right--proportional to the actual sensor resolution. Plus, take into account the fact that TrackIR is made in the USA (with much higher labor costs), and I think you're closing in on the rationale behind the gap.

[Winder]

Good post although touch screen's are expensive you've given me a great idea. Touchscreens use an array of touch switches which would eliminate any risk of damage to your expensive LCD screen. With a 1.5v watch battery you could run a touch switch and a IR LED in something the size of an elastic sewing thimble which would fit neatly on the end of your finger.

 

That's a good idea... I think my wife would give me way less grief if I tried something subtle like this, instead of a full on repurposed golfing glove. Unless it was an actual thimble, in which case I might as well present her my manhood on a silver platter. ;)

[TheMoose]

Big toys for big boys ;)

[Quarrion]

What part of 1024x768 do you not understand 'Winder? [EDITED by mod: unnecessary comment]

 

With bright tracking points you can get the full 1024x768 out of the wiimote, down to the pixel, actual sensor resolution is academic. With regards to frame rate, if you can't run Black Shark at more than 30fps, having a tracking device running much faster than that is also academic. Wiimotes are very common and so too are bluetooth dongles so you could already have them, or know someone who does, making the total cost $0.

Edited December 24, 2008 by Acedy

[Ramstein]

careful playing withy your Wii:megalol:

 

http://www.foxnews.com/story/0,2933,471364,00.html

[Acedy]

Fellas, stop the TrackIR vs. FreeTrack debate and get back on topic!

[Vault]

That's a good idea... I think my wife would give me way less grief if I tried something subtle like this' date=' instead of a full on repurposed golfing glove. Unless it was an [i']actual[/i] thimble, in which case I might as well present her my manhood on a silver platter. ;)

 

Hahaha not a nice flowery thimble. I'm thinking of using a thimble like this with a touch switch located on the interiour of the fingertip. Just a tip for refrence to anybody thinking of building a "finger tracker" the Wiimote tracks 940nm + IR LED's best. With a bit of careful work I think I can mount all the electronics in the interiour of this Thimble.

 

Quarrion I'm very, very happy with FT 2.2 and Wiimote. :thumbup:

[graywo1fg]

wow i cant wait to try this!!! going to pick all the stuff up sometime this Christmas break :D

[Winder]

What part of 1024x768 do you not understand 'Winder? [EDITED by mod: unnecessary comment]

I understand that every credible source on the Web that I know of disputes that number as the actual resolution for the Wiimote, including the lead developer for FreeTrack as well as the most prominent sources on Wii Homebrew. I understand that I've provided sources to back this up.

 

With regards to frame rate, if you can't run Black Shark at more than 30fps, having a tracking device running much faster than that is also academic.

It's certainly much more than an "academic" detail. First, having a camera that refreshes at the same rate (or slower, for that matter) that the game refreshes is a problem. It needs to refresh at least twice as quickly. Why? Phase drift. It is inescapable, and it ensures that the camera will never perfectly update in sync with the video card/game. To account for this, you need multiple frames of motion capture to be available for each frame of the game. That way, even if it's not balls-on accurate, when the game updates the head tracking data, it will be very, very close. If you don't have that extra redundancy, when phase drift kicks in, actual real time updating becomes impossible. This is guaranteed to happen with any camera that has a low frame rate. This is why web cams and other low FPS cameras have motion blur, latency, and fidelity issues.

 

Second, your understanding of FPS as it relates to motion tracking compares apples and oranges. Did you read my example about tracking a golf club? It's a pretty simple concept, I think--that being the difference between FPS as it pertains to an image/monitor/game's refresh rate versus a camera's refresh rate (especially when motion tracking is involved). In case you missed it, from a previous post:

 

"Another way to look at it… imagine you’re watching video of a golfer practicing his swing. Let’s say his club head is being tracked by markers using a motion capture system. You’re watching the results on a monitor that’s refreshing at 35 FPS (5 FPS faster than television, I believe). First, his club head is tracked by cameras that update at 15 FPS. Then his club head is tracked by cameras that update at 400 FPS. In both instances, the monitor you’re watching on is refreshing at the same rate. But the tracking quality is going to be vastly different. Why is this detectable to the human eye? Because the FPS is not impacting what you see from your monitor. The FPS is impacting what the monitor receives. This concept is exactly the same as what happens with head tracking. Cameras w/ low FPS don’t accurately detect motion real time, which means that by the time your see the in-game view with your eyes, what you see is already outdated."

 

Fellas, stop the TrackIR vs. FreeTrack debate and get back on topic!

OK, OK, noted. :music_whistling:

 

TBH, I don't think this discussion has been about TrackIR or FreeTrack, really. Maybe on an ancillary level, but the bulk of the discussion has just been some spirited disagreement about the fidelity of a Wiimote-based tracking system--which I think is relavent, considering the nature of the OP's hardware setup. Nevertheless, consider my jets cooled, following this post. :pilotfly:

 

Hahaha not a nice flowery thimble. I'm thinking of using a thimble like this with a touch switch located on the interiour of the fingertip. Just a tip for refrence to anybody thinking of building a "finger tracker" the Wiimote tracks 940nm + IR LED's best. With a bit of careful work I think I can mount all the electronics in the interiour of this Thimble.

 

 

I was thinking of something a little bit more masculine:

 

 

;)

[therion_prime]

It works amazingly well.

The microswich is inside the fingertip.

[Shrubbo]

Just should not be page 2 yet ;)

[Quarrion]

I understand that every credible source on the Web that I know of disputes that number as the actual resolution for the Wiimote

 

There is no indication of any dispute, the Wiimote's reported resolution has been well known for the last few years now.

'Each location is encoded in 10 bits and has a range of 0-1023 for the X dimension, and 0-767 for the Y dimension.' -- http://www.wiibrew.org/wiki/Wiimote

 

With regards to frame rate, if you can't run Black Shark at more than 30fps, having a tracking device running much faster than that is also academic.

 

It's certainly much more than an "academic" detail. First, having a camera that refreshes at the same rate (or slower, for that matter) that the game refreshes is a problem. It needs to refresh at least twice as quickly.

 

Point taken, but I mentioned Black Shark running at 30fps, the Wiimote, operating at 100fps, refreshes at least twice as fast, satisfying your requirement.

Edited December 25, 2008 by Quarrion

[Guest saied]

can someone tell me, what is this switch called that he uses. i cant find what it is called

[graywo1fg]

ALRIGHT!!! :D i got the stuff :P i think my biggest problem will be figuring out how to make my trackIR not interfere with my wiimote.... cant wait to see how it works!

 

Saied its just a "toggle switch or push switch" a Single Pull single throw or a SPST :D

Edited December 27, 2008 by graywo1fg

[JonK]

Outstanding!

 

Looking forward to hearing the results!

[Macka]

A good source of the parts is an old IR mouse you may have laying around. The mouse micro switches are perfect and of course you have either one or two IR led's inside as well. The USB cable has all the wire you need to join all together.

I use mine with one as a thumb switch on the side of the glove. (cotton lined rubber glove finger, reversed...Cheep....cheep!!!!....and nicely cleans the monitor as I fly.....lol.)

 

Great work JonK. :) :)

[JG-1_Vogel]

Wow, thats amazing. I'll agree with other statements here that this is just as revolutionary for simming as TrackIR was, I can see people who start using the "power glove" not being able to fly with out it after a while... I can't fly without Track IR now!

 

Seriously look into getting this done as a full on product, I'm sure there are companies out there who would be interested in taking on this as a commercial product!