Lichtknarre: Unmodified Wii remote as a sight accurate Lightgun using 2/4 LEDs

[RandyT]

I created this thread to share thoughts and experiences with this software, so as not to rough-up other threads.

It appears there is a piece of software called Lichtknarre that uses 4 IR LED tracking with unmodified Wii remotes.  Looks kinda janky, but I am going to try it.
Hopefully I have some bare IR LEDs somewhere.  Might try it with the dolphin bar first just to see how the software functions.

Thank you for posting about this software.  I went digging for just this type of application,but didn't see it referenced anywhere.

IMHO, this is one of the best ways to approach using the sensor in the Wii remotes for light guns.  Not only are there a number of gun shells commercially available for the controllers, but they are wireless and there are also dedicated guns which may end up working with the software, even if not using the same IR camera module.  Even cheap Knock-off Wii controllers may end up working with it, albeit likely not as well.  Based on my research, the important functions of the WiiMote IR camera module are exposed to the Wii through the remote, including the ability to sense up to 4 "blobs" (IR beacons.)  This means that anything which is possible to do with other solutions using those sensors as a foundation, should also be possible to do in an application like this one.  All without custom PCBs, desoldering components, building circuits, gutting expensive lightguns for their plastic shells, etc.

Definitely post the results of your tests here, if you would.  I will be digging the controllers out of my boxed up and stored Wii over the next few days to play with this and will probably do the same

Oof:  Probably should add a link so others won't need to hunt it down: Lichtknarre 2/4 LEDs Lightgun System
And by Badmouth's request, here is a link to the author's Github page for the Plugins, in case the smart folks here want to give it a shot: Lichtknarre Plugin

[ryoken]

 I try it. And I most say. The tracking and accuracy is really good. I also have the gun4ir and compare to that system the latency is almost the same. But it makes everything way more convenient because it can work on the normal wiimote. And is wireless

[RandyT]

I try it. And I most say. The tracking and accuracy is really good. I also have the gun4ir and compare to that system the latency is almost the same. But it makes everything way more convenient because it can work on the normal wiimote. And is wireless

Sounds very promising!  Thanks for the report.

[BadMouth]

I agree a dedicated thread is needed.  I haven't got around to testing it yet because I am dedicating a lot of time to other projects.  The author has the same guns as me though.     It's good to know that they will work ahead of time.  How well is yet to be determined.

[BadMouth]

I found a bag of LEDs with 940IR scribbled on it.  No idea of the specs and haven't messed with LEDs in a while, but a youtube video says 15mA to be safe, 20mA max.
Hooked up a 5v power supply (probably too much) and swapped out resistors until I got up to 19mA.  Nothing got hot.  They (or maybe the power supply) lasted about 20 minutes.

I made it far enough to verify that the guns connect and see four LEDs, but didn't make it to the point of aiming.  I could redo it using these LEDs with higher value resistors or run them in series, but the beam on them is so tight that they have to be pointed right at the gun. 

Not wanting to wait for AliExpress, I ordered what I could find on Amazon without spending too much:
3w 850nm IR LEDs with 140 degree lens built in
a constant current buck converter

To keep the cost to a minimum until I know it works, I didn't buy the pcb and housing they recommend.
For a heatsink I plan attach them to a short piece of round aluminum stock with thermal glue.
If the aluminum gets too hot I can make it longer or add heatsinks to the other end.

So not much to report yet, but I am working on it.

I did think their way of handling LED placement was kind of neat.

[RandyT]

I finally went digging and found a sack of 100ma 940nm LEDS I picked up ages ago.  Managed to dig up the datasheet for them, but it's the typical Chinese datasheet which gives the MAX without really providing the typical operating current.  They show testing data for 20 and 50ma of forward current and make reference to pulsed operation elsewhere, so I have no idea what they are safe to run at.  But I have a bag of them, so I guess I'll jack one up and let it go for a few hours to see what happens.  Still haven't found the Wii remotes though

The calibration instructions are interesting.  I take it to mean that it wants a uniform distance from the screen edge on all sides, and in those linear locations.  Is that the way you read it?  If so, the language for disregarding the instructions and measuring seems a little overcomplicating.

[BadMouth]

I finally went digging and found a sack of 100ma 940nm LEDS I picked up ages ago.  Managed to dig up the datasheet for them, but it's the typical Chinese datasheet which gives the MAX without really providing the typical operating current.  They show testing data for 20 and 50ma of forward current and make reference to pulsed operation elsewhere, so I have no idea what they are safe to run at.  But I have a bag of them, so I guess I'll jack one up and let it go for a few hours to see what happens.  Still haven't found the Wii remotes though

The calibration instructions are interesting.  I take it to mean that it wants a uniform distance from the screen edge on all sides, and in those linear locations.  Is that the way you read it?  If so, the language for disregarding the instructions and measuring seems a little overcomplicating.

I thought the instructions sounded complicated at first, but after a while it sunk in.  All it is really saying is the top and bottom LEDs need to be in the middle horizontally (the line between them splitting the screen 50/50) and then the height of the side LEDs (measured from the bottom LED) should be 26% of the distance between the bottom and top LEDs, not necessarily the screen.

[RandyT]

I thought the instructions sounded complicated at first, but after a while it sunk in.  All it is really saying is the top and bottom LEDs need to be in the middle horizontally (the line between them splitting the screen 50/50) and then the height of the side LEDs (measured from the bottom LED) should be 26% of the distance between the bottom and top LEDs, not necessarily the screen.

Any progress?  I did manage to find a Wii Remote, a nunchuk controller and my trusty old Activision/Cabelas "shotgun".   May find a use for that thing yet.

One thing which seems promising, is that on my Win10 machine, I just held the buttons according to the directions and the software linked it right up to my integrated Bluetooth adapter no problem. The buttons seemed responsive, but that's about as far as I got without LEDs set up to test the rest.  If the ones you ordered do a good job, I'll probably follow suit.

[BadMouth]

I made a wiring harness today for the LEDs and tested how warm they get.  They are 3 watt, but I think I can get away with no heatsink if I run them like 1 watt ones.  The $7 current limiter is definitely the way to go.  Adjustable and no messing with giant resistors getting hot.  I am running the LEDs in series at 6.5 volts and 300mA.  The package says they are rated for 700mA, but I am only giving them enough to get the job done.  Once they are in the cab behind smoked plexi, I may have to turn it up and add heatsinks.  I used fairly thick wire, so that is probably helping wick heat away.

The wide angle lens incorporated into the LED is also definitely the way to go.  The old LEDs were only visible straight on.  These are good from any angle and it's a very small package without the big lens and housing recommended by the author of the software.  I will post links when I get back to a PC.

My ancient bluetooth adapter wouldn't stay connected to both guns just like the old glovepie days, so I bought a new Asus one.  The first gun connects easy, but the second one takes multiple tries.  Once connected they stay connected, but the process has to be repeated every time the software is launched.  I might have just not gone far enough yet, but I don't see how this could be incorporated into a front end.

All I have to do is tape the LEDs to the monitor and I'll be ready for testing.  Might get to it tonight.  If not tonight, tomorrow.

[Locke141]

I haven't been on in a while but love this post. There's some YouTube videos up

[Fusselkroete]

Hi, this tool can also help you to configure the axis in the emulator.
You can find it for example under P1 > more Settings > Movement execution tool. This tool will than work for p1 as long it stays open. You can use the equivalents for the other players aswell.

https://drive.google.com/file/d/1Iqqrd5z-YkHZMC3wM-BR5rOcH67aZPtk/view?usp=sharing

[RandyT]

I could be wrong, but think the author just found us.  Probably due to the new traffic

[BadMouth]

I'll repeat a bit of my last post to pack more relevant info into this one.

LEDs are 850nm 3w with built a in 140 degree lens.  When they are powered, there is a bit of red visible to the naked eye.  It is subtle and not distracting, but 940nm ones would probably work just as well and have no visible light.  I like that I can see that they are on.
These are the ones I used: https://smile.amazon.com/dp/B08XQDLS8L?ref=ppx_yo2ov_dt_b_product_details&th=1 
I am not saying they are the best for the job.  I primarily bought them because I could get them quickly.

They were ran in series 6.5 volts and 300mA.  This is less than half of the rated current, but it keeps them cool enough to not use heatsinks.
This is the buck converter/current limiter I am using: https://smile.amazon.com/dp/B07G446KHJ?psc=1&ref=ppx_yo2ov_dt_b_product_details
(Not sure if it is a good one or not.  It took a while to get the voltage adjustment working.)

For the sake of completeness, I am using an Asus USBBT500 bluetooth dongle.  I am sure there are cheaper ones that will work.

This video was helpful in my understanding of how to set the LEDs up:


I just wired them in series in a loop (+ in then - to +, - to +, etc ).  (That is a thermocouple taped to the first one to measure temps)


It could just be that I'm inexperienced with it, but I found the calibration screen difficult to do holding the gun by hand so I ended up holding it with a phone mount on a tripod.
With the gun stationary, I could move the LEDs to get the correct numbers on the screen.  The side ones ended up needing to be moved inward closer to the edge of the screen while the top one needed moved off the top of the monitor.  I am sure this will be easier once I wrap my mind around everything, but it felt a bit overhelming trying to get it right at first.  I am sure I will be able to explain it better after doing it again on my cab.



Once the LED locations were calibrated I tested the gun as a mouse.  It was drifting a bit toward the edges.  I played around with the offset calibration to correct this.
I thought the offset calibration wizard was going to be helpful as it has you get close enough to only see two LEDs, but in the end I didn't find it useful and preferred to play around entering my own numbers.  Basically playing the high low game until I found the value where the mouse pointer was dead on with the gun sights near the edges of the screen.

So gun calibrated I launch MAME and learned how much I've forgotten.  No fake mouse input in MAME.  So I switch to the vjoy plugin and it doesn't work because I don't have vjoy installed on this PC.  So after getting vjoy and MAME square away, the first thing I notice is that the on-screen crosshairs in MAME trail pretty bad.  It doesn't cost me to miss any shots, but not ideal I guess.  Luckily there is a setting for this in Lichtknarre.  Under "More Settings" next to each player is a "Step" setting.  I double it and the trailing is pretty much gone.  Not sure at what expense, but the games were perfectly playable without crosshairs at the stock setting.

In mame with vjoy, the offset near the edges still seemed a bit off.  I am pretty sure it could be adjusted or corrected in the games original service menu, but I didn't want to get into that yet.  Using the sights on the gun, the shots were close enough to count as a hit in all the games I tried, but it still felt like it could use some tuning. 

I guess I should be more positive here:  Yes, it works well enough that you can use the gun sights and turn off the on-screen crosshairs..and from a distance I would expect to play light gun games at.  The gun was around 40" from a 27" monitor.   At least that is my experience with these Walther replica Wii guns.   

Vjoy is an extra layer that affects the testing, so I moved on to Demul where I could just use it as a mouse.

I started off with House of the Dead 2's service menu.  After playing some HOTD2 and Maze of Kings, I moved on to Sport Shooting USA.
This is the game to test guns on!  I took my time to dial in the calibration in the game's original service menu.
How well do they work?  Watch the video below.  I am aiming using only the gun's sights from 40 inches away.  You can see that the shot doesn't always correspond to the mouse pointer.  That is the game's original arcade calibration at work.  For the second round of shooting, I managed to get some decent footage from behind the gun.  The next to the last shot misses and you can see that I had the gun turned.  When I straighten it, the shot lands.  EDIT:  I also want to mention that the camera on my phone messed with the wii camera despite having the light disabled.  It would make the crosshairs jump to random places.  I put tape over the LED on the phone and the crosshairs became stable.


So in summary:

It meets my standards for accuracy and distance from the cabinet.
I will be installing this on my cab.
I will have some tuning to do to use the vjoy plugin.
Setup and calibration feels a bit klunky and stressful, but I am sure we can improve it once we understand it better.

Front-end integration is probably not going to happen, but I wouldn't rule it out completely.
It took a while to get both guns connected, but when I relaunched the program and pressed the sync button on them, they both immediately paired.
On these guns, it seems that I only have to press the sync button, not hold it down.  I think holding it down before was actually preventing them from pairing.

If you have wii guns, it is definitely worth it.

EDIT: Randy, please add the author's github to the first post: https://github.com/Geekonarium/LichtknarrePlugin
Maybe one of the talented people at BYOAC will take an interest in helping develop it.

[RandyT]

Plugin page link added to the original post.

Very nice writeup.  But a couple of comments/questions. 

(Edit: Just re-read your post and it wasn't clear whether the supply was 6.5v or whether that is what you had it adjusted to through the buck converter.  If the latter, you can disregard the next paragraph.  Probably still worth reading, as a 6.5v power supply might be a viable alternate approach and I have no idea what you are feeding the converter.)

The first is: Is the buck converter even necessary with 4 of those LEDs in series and the 6.5v power supply, assuming that is what you are using?  Based on my shoddy calculations (so take them with a grain of salt), if each of the LEDs are being driven at 1.6v (roughly 1/4 of the available voltage) with a draw of 300ma, only 30mw of power would need to be dissipated by a .33ohm resistor, so even a 1/4 watt resistor would be overkill.  I think the difficulties you are having with the buck converter is that it is a step-down converter, and there may not be enough differential between input and output for it to work well. Hell, with a decent 6.5v regulated supply, I'd just hook the 4 of them in series, without a resistor or anything, just to see what happens.  But that's me, so check the numbers and make your own decisions. 

And as usual with the Chinese datasheets, the data they provide seems to be contradictory...at least for the two longer wavelength IR parts.  The graph shows a voltage range of 1.8 to 2.2, while the ratings chart states 1.6 to 2.0.  If I were to guess, they just copied the graphs from the 740nm parts, so they aren't correct for the other IR parts.  The actual curves probably won't differ too much, so you can probably just drop them down a couple of points on the voltage scale to get a ball-park and stick to the 1.6v to 1.8v rating.

About your setup.  Visually, and based on the position of the blobs, I think your LED positioning was at least initially incorrect.  As I stated earlier about the calibration explanation, assuming your monitor image fits perfectly to the frame, my gut feeling is that the LED's need to be an equal distance from the edge on all 4 sides, and at the positions indicated by the positioning calibration lines image you posted earlier.  From your photo, the top and bottom LEDs are outside the frame, and the left and right are inside, with one of them appearing to be closer to the screen edge than the other.  Those lines connecting them should be a nice, straight cross pattern, and I believe that what it may be showing you is the positioning error.  Even though you got it to work, I think the position of the LEDs made it a lot harder than it should have been.  Again, just an observation, I wasn't there.

And one final comment is the speakers.  Maybe it's not a problem, but with the wide angle of the LEDs, I have to believe that at some angles, you are going to get reflections off the sides of the speakers which will impact the accuracy of the system.  Depending on how the code is written, this can affect not only stability of the cursor, but possibly also tracking speed, as some of the jittery values caused by this could be seen to be "out-of-reasonable-range" and discarded as error, causing the system to be waiting for something more reasonable to arrive to be averaged into the position data.  To get a good test, I would think that at minimum, the screen should be brought out past the front of those speakers.

It might also be interesting to try with an OG Wii remote, as the camera in the gun, as nice (and rare) as that gun is, may be inferior to the ones used in the OEM Wii remotes.  If it's the one with the sub-board with a bunch of components around the camera and a couple of black blobs on the back, then it's probably the same as the Chinese "knock-off" Wii remotes, which reportedly aren't very good.

Ok, I'm done for now 

Edit:  Ok, one more thing...the reason your phone was interfering was likely due to the auto-focus.  Probably an IR range finder/lidar thing sending out pulses when the camera is on.  If you can turn off auto-focus, it might stop doing that

[BadMouth]

I am stepping down from 12 volts.  My cab has a 12v power supply in it for the monitor rotation that I plan to tap into.  They will eventually be turned on and off via software.  I also might need to up the current to make them brighter behind the smoked plexi on the cab.

The package the LEDs came in said 1.5-1.8V .700mA (which I assume is 700mA, so who knows if these are correct.)

I started with the LEDs an equal distance from the edge of the screen.  Where they are now is where they ended up through the calibration process.  It's all about their relationship to each other so all four could be moved down, but the sides would still be right on the edge of the screen while the top and bottom aren't.  Getting it this good was not easy.  You will see, lol.  Hopefully the cab will come out better since I have some experience now.

The speakers are behind the front edge of the monitor.  Pretty sure what you are seeing is lens flares.  I was wondering if they were as bad for the Wii camera.

This was just a temporary taped on setup to see if it worked well enough to bother with.  It does.

EDIT:  ahhhh.....the autofocus.  Makes perfect sense.

[BadMouth]

I also forgot to mention that with vjoy the games have to be played stretched widescreen.  I think calibrating in the in game service menu on games that allow for it could allow them to be played in 4:3,  but haven't tried it.  Also remember to always decrease MAME's default 30% deadzone and increase saturation to 100% for analog controls.  It just dawned on the that the trailing on screen crosshairs probably could have been improved by cranking up the sensitivity in MAME's in game options.  I wasn't thinking in terms of analog controls, but if it is ran through vjoy and showing up as joysticks, all those settings come into effect.

[RandyT]

Everything sounds good then.  I noted the lens flares, and since the Wii camera has a simple, single focus lens, I don't think it experiences them.  I swear the photo makes it look like the monitor was inset, but photos are funny things when it comes to finding depth.  And there should be no problem stepping down from 12v.  Maybe just a janky multi-turn pot.  Apologies for the back-seat driving, just thinking extremely out-loud   At least others reading my ramblings might see some things to avoid.

I was still trying to get my brain to swallow the LED positioning (sorry for being a dingbat).  But I think I get it now.  The horizontal line of the on-screen LED position screen is only useful if you are able to position the top and bottom LEDs at exactly the points to which they intersect with their respective screen edges.  I think what needs to be done to simplify this is to make this a multi-step process with either some extra code or a small external app.  The way I see this working would be:

1: A vertical line is shown on-screen as a reference to center the top and bottom LEDs, maybe with some verbiage to place them as close as possible to the active screen edges to keep the shooting distance reasonable, due to the camera's limited vertical FOV.

2: Put up a couple of input boxes for the user to enter the measured distance from the center of the LED or cluster to the active screen edge for top and bottom respectively.  Then from those two values, calculate a vertical offset to achieve the proper ratio.  I have to think about this more to come up with the formula, but I don't think it's too complicated.

3: Show a horizontal line on the screen at the position calculated in step two, with some verbiage to place the side LEDs in-line with it and to maintain equal spacing from the active screen edge, which should keep them equidistant from center.

Does this sound remotely correct?

And thanks for the link to the LEDs.  Amazon's curated search results must have kept me from seeing those, as they never came up in my attempts to find them.   
Edit:  FWIW, it kept showing me these, which are cheaper and are probably the same, but probably wouldn't get delivered as fast as not shipped by Amazon.

[Fusselkroete]

Everything sounds good then.  I noted the lens flares, and since the Wii camera has a simple, single focus lens, I don't think it experiences them.  I swear the photo makes it look like the monitor was inset, but photos are funny things when it comes to finding depth.  And there should be no problem stepping down from 12v.  Maybe just a janky multi-turn pot.  Apologies for the back-seat driving, just thinking extremely out-loud   At least others reading my ramblings might see some things to avoid.

I was still trying to get my brain to swallow the LED positioning (sorry for being a dingbat).  But I think I get it now.  The horizontal line of the on-screen LED position screen is only useful if you are able to position the top and bottom LEDs at exactly the points to which they intersect with their respective screen edges.  I think what needs to be done to simplify this is to make this a multi-step process with either some extra code or a small external app.  The way I see this working would be:

1: A vertical line is shown on-screen as a reference to center the top and bottom LEDs, maybe with some verbiage to place them as close as possible to the active screen edges to keep the shooting distance reasonable, due to the camera's limited vertical FOV.

2: Put up a couple of input boxes for the user to enter the measured distance from the center of the LED or cluster to the active screen edge for top and bottom respectively.  Then from those two values, calculate a vertical offset to achieve the proper ratio.  I have to think about this more to come up with the formula, but I don't think it's too complicated.

3: Show a horizontal line on the screen at the position calculated in step two, with some verbiage to place the side LEDs in-line with it and to maintain equal spacing from the active screen edge, which should keep them equidistant from center.

Does this sound remotely correct?

And thanks for the link to the LEDs.  Amazon's curated search results must have kept me from seeing those, as they never came up in my attempts to find them.   
Edit:  FWIW, it kept showing me these, which are cheaper and are probably the same, but probably wouldn't get delivered as fast as not shipped by Amazon.

The white bars and the ratio number display should be the indicator so that the ratio is properly maintained. The problem here is that the program never knows where exactly the screen is. Whether there is more space on the left, right, top or bottom. The Wii simply cannot see this. Therefore, only the imbalance can be shown by the currently white lines.

http://forum.arcadecontrols.com/index.php?action=dlattach;topic=164992.0;attach=392263;image

It is also very difficult to calculate that the other way around, as multistep. If there would be a new screen where the program tries to calculate this the other way around, then the program does not know which ratio the screen has, if it would assume that the user has put the LEDs directly to the screen.


Also:

It is possible to have correct aiming also at the edges. Simply start to test at the left and increase/decrease the left offset in small steps and test inbetween. Continue with right offset. I will try to make a video about it in the near future.

https://geekonarium.de/en/what-is-ir-offset-and-why-do-i-need-it/

Also:

I heard for normal size screens 980nm IR LEDs works also fine!

Also:

Yes please beware reflections from walls or other objects a rim around the LED could help or only on one side.

[BadMouth]

When I add LEDs to my cab, I think I can work out steps for using the current calibration screen that is easier to explain and gets better results than my quick taped on test.
It might be a while before I get to it though.  I have some other things that need done first and I want to design and 3D print some brackets for the LEDs.

[BadMouth]

Fusselkroete, thank you for providing more information.   

Has there been much interest in Lichtknarre?  My reaction was "Why haven't I heard of this before?"
It works great!

[Fusselkroete]

Fusselkroete, thank you for providing more information.   

Has there been much interest in Lichtknarre?  My reaction was "Why haven't I heard of this before?"
It works great!

  

not sure how many ppl use this tool regular. no tracking.

Downloads seems to be not that bad for this kind of geeky tool
https://geekonarium.de/en/download-changelog-lightgun-lichtknarre-wiimote-line-of-sight-pc-windows/

[RandyT]

The white bars and the ratio number display should be the indicator so that the ratio is properly maintained. The problem here is that the program never knows where exactly the screen is. Whether there is more space on the left, right, top or bottom. The Wii simply cannot see this. Therefore, only the imbalance can be shown by the currently white lines.

http://forum.arcadecontrols.com/index.php?action=dlattach;topic=164992.0;attach=392263;image

It is also very difficult to calculate that the other way around, as multistep. If there would be a new screen where the program tries to calculate this the other way around, then the program does not know which ratio the screen has, if it would assume that the user has put the LEDs directly to the screen.

I understand what you are saying, and it has occurred to me that one more piece of information by the user would be required.  I'll regret this (because I'll probably be wrong), but I'll try to better explain how I think this can be accomplished.

If the user supplies the distance between each of the top and bottom LEDs to their respective edge of the active screen, and then also supplies the distance between the top and bottom LEDs, it would actually allow the LEDs to be mapped in relation to the screen as percentages of deviation.  Those percentages could then be applied to the vertical resolution of the calibration screen which shows where to place the side LEDs, creating a "virtual screen" upon which the correct placement ratio and line position can be calculated.  This virtual screen is then laid over the physical screen using the LED-to-edge deviation percentages to find the offset.

The following is an example using arbitrary units:

Edit: Original Example removed.  It was simpler than I thought.

Top LED Distance = 1
Bottom LED Distance = 2
Distance Between LEDS = 24

From this information, we know that the screen is 21 arbitrary units tall.  Therefore, for a 1080p image, the number of screen pixels per unit is 1080/21=51.42857142857143

We also know that there are 3 additional pixel units we need to add to account for the LED spacing, so our virtual vertical screen size becomes 1080+(51.42857142857143 * 3)=1234.285714285714 pixels

From there, we can find the pixel row at 74% with 1234.285714285714 * .74=913.3714285714286th row of pixels on the virtual screen.

And finally, to equate that to a row on the physical screen, we offset the starting point by the Top LED Distance * Pixels Per Unit, or in this case 51.42857142857143, so the line would be drawn at row 913.3714285714286-51.42857142857143=861.9428571428571 or 862 rounded up.

[Fusselkroete]

The white bars and the ratio number display should be the indicator so that the ratio is properly maintained. The problem here is that the program never knows where exactly the screen is. Whether there is more space on the left, right, top or bottom. The Wii simply cannot see this. Therefore, only the imbalance can be shown by the currently white lines.

http://forum.arcadecontrols.com/index.php?action=dlattach;topic=164992.0;attach=392263;image

It is also very difficult to calculate that the other way around, as multistep. If there would be a new screen where the program tries to calculate this the other way around, then the program does not know which ratio the screen has, if it would assume that the user has put the LEDs directly to the screen.

I understand what you are saying, and it has occurred to me that one more piece of information by the user would be required.  I'll regret this (because I'll probably be wrong), but I'll try to better explain how I think this can be accomplished.

If the user supplies the distance between each of the top and bottom LEDs to their respective edge of the active screen, and then also supplies the distance between the top and bottom LEDs, it would actually allow the LEDs to be mapped in relation to the screen as percentages of deviation.  Those percentages could then be applied to the vertical resolution of the calibration screen which shows where to place the side LEDs, creating a "virtual screen" upon which the correct placement ratio and line position can be calculated.  This virtual screen is then laid over the physical screen using the LED-to-edge deviation percentages to find the offset.

The following is an example using arbitrary units:

--------------------
Top LED Distance= 1
Bottom LED Distance= 2
Distance Between LEDs = 24

From this information, we now know that the active screen height is 21 and the upper deviation is 1/24 or 4.1% and the lower deviation is 1/12 or 8.3% of the total distance. 

So if we are using a 1080p resolution screen to show side LED placement, we use the figures above to create a virtual screen which is 4.1% larger at the top and 8.3% larger at the bottom. 

The vertical height of our virtual screen is 1080+((1080*.041)+(1080*.083))=1213.92 pixels

Therefore, the indicator line for the LEDs would be placed at the 1213.92*.74=898.30th row of pixels on our virtual screen.

Then all we need to do is draw the line for LED placement on our physical screen at the 898.30-(1080*.041)=854.02th row of pixels. 
-------------------

Please double check my results.  Even if it's correct, there are probably some shortcuts to get to the same result.  I just wanted to step through it to show the process for getting there.  Obviously, the measurements supplied by the user would need to be accurate for it to be perfect, but if they are close, it could aid in providing a starting point.

The problem with that theory is that you can't work with physical and virtual units. pixels has no reference to real life. A virtual screen with 1024x768 for example could be 4:3 or 16:9. the resolution does not care. But theoretical a screen with all exact physical messured values (width/height of the screen + offsets left/right/top/bottom) could be implemented, but i think it's too much for the user. (the missed point look update)

Whats the problem with the actual screen? Get the ratios right it says More exact you do it better it the aiming is Try to satisfy the values as best as you can!

Atm user only struggles with IR-Offset.
The offset wizard aims to solve this problems. But it's hard to aim perfect and maybe there is still a little bug in that wizard. Need to think about it to improve it.

The offset wizard knows that the dot which the user is aiming for left offset is for example 25% of the screen. So when you aim the dot and shoot in that moment the app knows where the left IR and the right IR is. assumed that this was 100% and can now see how much percent it has as offset for the left side. I think here a bug could be that it doesn't know the middle point of the screen if i remember right. so if left has different distance to right. will check on that.

Also thinking about to use the mouse as a helper in the offset wizard soon. Because the mouse is the end goal. Than the user can press some buttons on the wii and iterate through the offsets in small steps in this wizard. Maybe that will improve it. To get the offset right is more a problem of the interface atm i think, to make it practical.

Will see how to improve.

Thanks for you'r feedback and the feedback of the others! This forum realy helps to see what users struggles hard with

Update:

Sorry i missed the point that you want the user to provide all as distances ir-offsets/ir-width(distance from left IR to right IR)/ir-heights(distance from top-IR to bottom-IR) for example in mm. Was too early in the morning . ok maybe that will work, but i think an intuitive calibration process without letting the user measure stuff is much better? Aiming and showing whats wrong is much better imop. Will think about the advantages/disadvantages between these methods. Maybe this could be become part of offset wizard or something or an extra ir offset calculator. Thanks 4 feedback again.

Your other point was to have this as initial calibration step. As more do i think about this it could work. But i still think to let the user satisfy values on the fly is much more easy for the user. Because these are real values which comes from the wii remote and not from the ruler. Some ppl make mistakes when doing measurements and are confused the other way around . Sounds like both ways have advantages/disadvantages.

[RandyT]

Update:

Sorry i missed the point that you want the user to provide all as distances ir-offsets/ir-width(distance from left IR to right IR)/ir-heights(distance from top-IR to bottom-IR) for example in mm. Was too early in the morning . ok maybe that will work, but i think an intuitive calibration process without letting the user measure stuff is much better? Aiming and showing whats wrong is much better imop. Will think about the advantages/disadvantages between these methods. Maybe this could be become part of offset wizard or something or an extra ir offset calculator. Thanks 4 feedback again.

Your other point was to have this as initial calibration step. As more do i think about this it could work. But i still think to let the user satisfy values on the fly is much more easy for the user. Because these are real values which comes from the wii remote and not from the ruler. Some ppl make mistakes when doing measurements and are confused the other way around . Sounds like both ways have advantages/disadvantages.

No problem.  I get "false starts" when I think I understand something all of the time (see earlier)

The user is going to need to measure stuff for that initial LED placement regardless.  As the center LED is not in-line with the sides, it's currently very difficult to start out close to the having the side LEDs at the correct locations.  To do this well, the user would need a ruler, something to draw lines on the TV frame, make multiple measurements and reference marks, etc.

My belief is that actual references provided on-screen, like the one you have showing the optimal placements at the screen border intersections, will always be the easiest for the user to follow and negate the need for the extra stuff.  If it works out, it should also be more accurate.

I'm not advocating removal of any fine-tuning screens.  This is just a way to get close right from the start without needing 3 hands and a calculator to get there .

Edit:  The user shouldn't need to provide measurements for everything.  Just the 3 measurements I noted.  Unless the active screen is not centered, the user should be able to "eyeball" the left to right dimensions of the side LEDs and place them close to where appropriate, so long as there is a line on the screen to follow for the vertical placement.  Also, the actual unit of measure is arbitrary.  It can be mm, inches, soup spoons, etc...  so long as the same unit of measure is used.  It is just used for finding the percentage of deviation with whatever the unit is that is currently being used.

[Fusselkroete]

Here an idea how your calibration process could look like:

Step1: User places top LED 50/50 horizontal. An arrow is shown on the screen for the 50/50 top edge.
Step2: User places bottom LED 50/50 horizontal. An arrow is shown on the screen for the 50/50 bottom edge.
Step3: let the user enter top/bottom offsets and distance between the 2 LEDs.

Step4: programm calculates with the offset the screen top position for the left and right IRs. Because it knows the offsets and heights of the screen. it now can calc the percentage for the screen and apply it on the resolution.
Step5: Let the user place Left/Right Leds according to the edge positions which was calculated in step4.
Step6: let user enter Left/right offsets.

Hmmm but to be honest i bet this still do not work. See all the stuff directly over the wii camera is the best because it translate directly to the calculation. Thats why i prefer a more try and error process over the wii cam. Like it is now. Satisfy the values by moving the leds. After this is done adjust the mouse by trying out more and less offset. The offset wizard could be improved alot to this workflow.

Lets call my process the wii camera calibration process and the other one the manual calibration process:

I bet if you only one mm with one thing off with the manual calibration process than it doesn't work.

Also you do not need marks for moving the leds with camera. Maybe glue strips or something. I use magnets to have it a little bit variable.

I should do a tutorial video how to calibrate and show it in front of the calibration process Will see maybe i can find a hybrid of both worlds.

To summarize:
- Manual process needs a ruler
- Wii cam process needs to replace the LEDs.

There must be a process from both worlds to eliminate both disadvantages

Step1: Like let the wii see the leds the whole time.
Step2: User places top LED 50/50 horizontal. An arrow is shown on the screen for the 50/50 top edge.
Step3: User places bottom LED 50/50 horizontal. An arrow is shown on the screen for the 50/50 bottom edge.
Step4: Let the user shoot two dots on the screen to find the offsets. One 10% top of the screen and one 10% bottom of the screen 50/50 horizontal.

Step5: programm calculates with the offset the screen top position for the left and right IRs. Because it knows the offsets and heights of the screen. it now can calc the percentage for the screen and apply it on the resolution.
Step6: Let the user place Left/Right Leds according to the edge positions which was calculated in step4.
Step7: Let the user shoot another two dots on the screen for left and right offset.

But shooting dots works that well as the actual IR offset wizard works haha. And now its a long process where the user is more irretated. Not so sure.

Even though we are now theoretically thinking about a better calibration process, which is also super nice and helps me a lot, but I will perfect the wii camera try and error process. A better interface for the offset wizard and a tutorial video what is shown before calibrating to make it more practical. sticky tape, how try and error with IR offsets and what it all means.

Also this is all one time calibration.

[RandyT]

Here an idea how your calibration process could look like:

Programatically, yes, for the most part.

I see it like this from a user perspective:

Step1: Vertical center line shown to user for top and bottom placement.  User instructed to place LEDs in-line and as close to screen edge as is practical.

Step2: User is prompted to enter top/bottom offsets and distance between the 2 LEDs, with emphasis on accuracy.  The more accurate the better.  No rounding.

Step3: Program calculates the correct vertical offset for the left and right IR LEDs and places a horizontal line on-screen at that location.  User instructed to place LEDs in-line and as close to screen edge as is practical, keeping the same distance between the LED and the active screen edge on both sides.

From that point on, it should just be fine-tuning by whatever means is most practical.  I haven't gone far enough in the setup process to see exactly what that entails.  But if you already have the ability for the user to enter offsets for the LEDs, and the Wii remote camera has the ability to see them, I can envision using the information from the setup process to place a representation of the physical LED pattern on the screen and then instructing the user to position themselves and the "camera" to try to align the blobs seen by the camera as close as possible with the ones on the screen and pulling the trigger.  At that point, the software could automatically calculate the offsets between each.  If anything still needed tweaking, you would still have the ability to do so.

Maybe that last part wouldn't work for how you are doing things, but it's something I would personally try as an experiment.

[Fusselkroete]

But many ppl had no problem and managed to master the calibration process and have perfect aiming. Its only one time calibration so all is good

[BadMouth]

Randy, maybe reserve judgement until you've actually set is up and used the calibration.  It could be that I just didn't wrap my head around it completely and made it out to be harder than it is.

[RandyT]

Randy, maybe reserve judgement until you've actually set is up and used the calibration.  It could be that I just didn't wrap my head around it completely and made it out to be harder than it is.

Yeah, I'm a "fixer" by nature and it looked like there were some aspects you encountered which might have benefitted from some input.  But in the end, it's a process which only need be done once.  So as long as the system works well, which it appears to, that's all that's really important.

Just so my comments are not misunderstood, Fusselkroete's work on this is greatly appreciated.  I think it will make good lightgun gaming far more accessible to gamers than possibly anything has in the last decade or so.  IMHO, it deserves more attention than it appears to have received so far.  Hopefully that situation changes as more start using it and the "word-of-mouth" kicks in, which will hopefully encourage him to take the software further.

As a side note, my Nyko "Perfect Shot" gun shells showed up today.  They feel great in the hand, are not bulky at the forward part and the trigger action is excellent.  They even have a port pass-through extension which brings it down below the grip.  I have another style on order as well, but the Nyko's will be tough to beat.  My only complaint with them is the lack of a bead on the clip in front, which you really can't fault them for.  The Wii never really had good line-of-sight aiming, so it wasn't really necessary.  I think a small, well-placed divot, a BB and a drop of superglue will take care of that nicely.  Now just waiting on the LED modules    

[RandyT]

I couldn't help myself.  Based on my earlier ramblings, I made this small app for myself to try when the time comes.  It might work, or I might have just wasted my time.   

BadMouth, if you feel like it, go ahead and check this little program out and see if the result points to where your LEDs are stationed.  If not, then we'll know it's useless and I'll delete it.  Otherwise, it may help a bit when it comes time to move your LEDs to your cabinet.  It's resolution agnostic, but should at least be used at the native aspect ratio of the display.  Also, at low resolution some of the screen elements might trample each other.  Outside of that, it's been tested and based on the test measurements, it looks like it does what it should.

*edit* Software used some bad math.  Will update in a later post. 

[BadMouth]

Installed LEDs on the cab today.  I had all kinds of ideas in my head for holders, brackets, heat sinks etc.  Due to time and space constraints I ended up using black duct tape. 
I wanted all of the LEDs at the edge of the screen to get the tightest grouping possible, allowing the guns to work at the shortest distance possible.

So here is what I did this round....
> Placed top and bottom LEDs on the edge of the screen centered by measurement.  This corresponded with the vertical line in "2. default LED calibration"
> Placed left and right LEDs on the ege of the screen based "3. default gun calibration" making sure that they were both the same distance from the bottom of the screen and the line between them was straight.  This resulted in them being a bit below the horizontal line in "2. default LED calibration". 

This is why the top LED ended up high on my previous try.  It had to be raised to get the right ratio.  This time the side LEDs were moved down to get the correct ratio.  This way is more correct because it keeps all the LEDs on the edge of (or same distance from) the screen.



Fusselkroete,
For "3. default gun calibration", I need to hold the gun right of center or turn it to get the circles in the boxes and lines straight.  Both guns are the same way.


Is it better to keep the gun perpendicular to the screen and shift it to the right, or is it better to keep the gun centered with the screen and turn it?  Either way the mouse pointer ends up shifted left of where the gun is pointing.  I was able to get it close with the following offsets:


Am I using the offsets correctly or is there something else incorrect before applying the offsets?
I could shift the top and bottom LEDs, but that seems way incorrect because then they wouldn't be centered.

EDIT: Typed this before/during Randy's post.....

[Fusselkroete]

Installed LEDs on the cab today.  I had all kinds of ideas in my head for holders, brackets, heat sinks etc.  Due to time and space constraints I ended up using black duct tape. 
I wanted all of the LEDs at the edge of the screen to get the tightest grouping possible, allowing the guns to work at the shortest distance possible.

...

@BadMouth
Try following: Put all offsets to zero again. Test aiming left side with mouse plugin. Increase/Decrese offset by 1 and aim the screen and test if it becomes better.  If you think you overshooted make smaller steps. Test till you satisfied. Than do the same with the right side. After your done try again on left and right. Use the same procedure for top/bottm. Your setup Looks like it has zero IR-Offset, from what i can see on the images. I dont see bottom IR.

[BadMouth]

I couldn't help myself.  Based on my earlier ramblings, I made this small app for myself to try when the time comes.  It might work, or I might have just wasted my time.   

BadMouth, if you feel like it, go ahead and check this little program out and see if the result points to where your LEDs are stationed.  If not, then we'll know it's useless and I'll delete it.  Otherwise, it may help a bit when it comes time to move your LEDs to your cabinet.  It's resolution agnostic, but should at least be used at the native aspect ratio of the display.  Also, at low resolution some of the screen elements might trample each other.  Outside of that, it's been tested and based on the test measurements, it looks like it does what it should.

If anyone else wants to run it, feel free.  But it's not too exciting

Randy,
Since the LEDs were already in place I used your program to see if the line would end up where my LEDs are.
It ended up above my LEDs by the amount of the top offset.
Top offset: 11mm, bottom offset: 11mm, distance between LED centers: 346mm

Not sure if it's an error in the program, but I can't move my LEDs very easily at the moment.

[Fusselkroete]

I couldn't help myself.  Based on my earlier ramblings, I made this small app for myself to try when the time comes.  It might work, or I might have just wasted my time.   

BadMouth, if you feel like it, go ahead and check this little program out and see if the result points to where your LEDs are stationed.  If not, then we'll know it's useless and I'll delete it.  Otherwise, it may help a bit when it comes time to move your LEDs to your cabinet.  It's resolution agnostic, but should at least be used at the native aspect ratio of the display.  Also, at low resolution some of the screen elements might trample each other.  Outside of that, it's been tested and based on the test measurements, it looks like it does what it should.

If anyone else wants to run it, feel free.  But it's not too exciting

Randy,
Since the LEDs were already in place I used your program to see if the line would end up where my LEDs are.
It ended up above my LEDs by the amount of the top offset.
Top offset: 11mm, bottom offset: 11mm, distance between LED centers: 346mm

Not sure if it's an error in the program, but I can't move my LEDs very easily at the moment.

Look example calculation. From cm to %.
https://geekonarium.de/en/what-is-ir-offset-and-why-do-i-need-it/#example1

But i have better results with try and error procedure.

[BadMouth]

LEDs are as close as I could get them to the screen without covering any of it.  That's why I was questioning the fairly high offsets. 

I redid again and ended up with -7 & +9 for left and right and -4 & +5 for top and bottom.  It is working good with those settings.
I just wanted to make sure that the need for the larger offsets wasn't caused by mistakes in the previous steps.

Either way, the gun tracks well after the offsets are applied. 

[Fusselkroete]

LEDs are as close as I could get them to the screen without covering any of it.  That's why I was questioning the fairly high offsets. 

I redid again and ended up with -7 & +9 for left and right and -4 & +5 for top and bottom.  It is working good with those settings.
I just wanted to make sure that the need for the larger offsets wasn't caused by mistakes in the previous steps.

Either way, the gun tracks well after the offsets are applied. 

Looks little bit strange these offsets But glad it works for you now. yeah this step is a bit fiddly but its one time setup ^^

[RandyT]

Not sure if it's an error in the program, but I can't move my LEDs very easily at the moment.

I'll look at my code to see if I did something dumb.  It should be ok, as the software came up with the same answers as the manual example I went through earlier.  But if that's not correct, it could be a case of "right answer to the wrong question"

[BadMouth]

Randy, just noticed my side LEDs are off by about the same amount in your program as the line in "2. default LED calibration" in Lichtknarre.

I used the  "3. default gun calibration" dynamic calibration screen to set the height of my LEDs.  Using the lines in "2. default LED calibration" resulted in the ratio being off and then having to move the top LED up, which I didn't want to do.
 
I don't want to jump to conclusions.  I'll wait to see what you experience when you set up and test.

Plenty of people may have used the program, but how many of them are old codgers who grew up with accurate CRT light guns.   

At any rate, it works perfectly with the offsets applied.

[RandyT]

Thanks for bearing with me.  My calculations were bad and I used it for the software without verifying.  Where's the self flagellation emogi?   

@BadMouth (and whoever else might be following along), please delete that first version and try this one. I think (hope) this is a keeper.

[BadMouth]

Thanks for bearing with me.  My calculations were bad and I used it for the software without verifying.  Where's the self flagellation emogi?   

@BadMouth (and whoever else might be following along), please delete that first version and try this one. I think (hope) this is a keeper.

Still a tad above, but since I have the oddness of larger offsets than expected, I wouldn't bother trying to revise it until you set your own up.  Might be something weird on my end. 

I still have to wire the power supply into the cab and transfer my gun gamelist from my old setup.  But as far as test screens and windows mouse control go, the guns are working as expected.