Wii accelerometer details

#51
Rumble will have a limited impact on accelorometers's drift I think , except on what's happening on screen.
Fishing example :
Wiimote rumble -> fishing rod will move accordingly to the rumble.

Rumble affect accelorometer but in a good way since if your hand vibrate, the cursor on screen will to making it hard to aim like in real life.
 

Gahiggidy

My aunt & uncle run a Mom & Pop store, "The Gamecube Hut", and sold 80k WiiU within minutes of opening.
#52
JoshuaJSlone said:
Hey guys, here's yet another company involved in some aspect of the controller, PixArt. This is a press release from yesterday.



I'm looking at PixArt's site, but it seems a bit crappy and out-of-date. It's not obvious to me which product is the one in the remote, or at least which is the closest.
Dude, I think you may have discovered that the WiiMote has a CAMERA inside it!

Most of thier stuff appears to be optical chips for camera-phones and laser mice. This must mean that the WiiMote's tip has a camera that is looking at the sensor bar to get a reading of its position in relation to the TV set.

Wow.

----

edit. Ok. Reading on it seems everyone else got this point too.

- Hey... here's a thought. If the WiiMote's speakers have a memory chip to store the sound effects, could that same memory chip be involved with storing digital pictures? (if indeed they use the CMOS as a camera).
 

Galactic Fork

A little fluff between the ears never did any harm...
#53
Gahiggidy said:
edit. Ok. Reading on it seems everyone else got this point too.

- Hey... here's a thought. If the WiiMote's speakers have a memory chip to store the sound effects, could that same memory chip be involved with storing digital pictures? (if indeed they use the CMOS as a camera).
I don't think the CMOS would be a camera in the complete sense. It only needs to detect one specific thing, so it doesn't even need anything that reads colors or even light and dark, it only needs whatever is special about the sensor bar. It may have a lens but I bet the actual receiver is designed specifically for the bar.
 
#54
The controller has an IR filter over the front of it so that it only sees IR light. Wouldn't make a very good camera.

Plus, it looks like it has some image processing built in so that it's only sending information about the points (LEDs) it sees; it's not sending back a whole image to the console; that would require too much communication (be slow and drain battery power).

With 2 LEDs in the sensor bar, the controller can tell what direction it is pointed with respect to the LEDs, and it can tell how far away it is from the bar (if it can see both LEDs). It cannot detect its own position in 3D space, though. You'd need a minimum of 3 non-collinear LEDs for that purpose (4 is better).
 
#55
Tekky said:
With 2 LEDs in the sensor bar, the controller can tell what direction it is pointed with respect to the LEDs, and it can tell how far away it is from the bar (if it can see both LEDs). It cannot detect its own position in 3D space, though. You'd need a minimum of 3 non-collinear LEDs for that purpose (4 is better).

isnt this essentially describing motion in 3D space? or an enough to complete a calculation through relevant data?

edit: i think i misread this then, you are saying here the bar alone can't detect its position yes?
 

Gahiggidy

My aunt & uncle run a Mom & Pop store, "The Gamecube Hut", and sold 80k WiiU within minutes of opening.
#56
Tekky said:
The controller has an IR filter over the front of it so that it only sees IR light. Wouldn't make a very good camera.

Plus, it looks like it has some image processing built in so that it's only sending information about the points (LEDs) it sees; it's not sending back a whole image to the console; that would require too much communication (be slow and drain battery power).

With 2 LEDs in the sensor bar, the controller can tell what direction it is pointed with respect to the LEDs, and it can tell how far away it is from the bar (if it can see both LEDs). It cannot detect its own position in 3D space, though. You'd need a minimum of 3 non-collinear LEDs for that purpose (4 is better).
Don't forget the FLAP. What's UNDER the FLAP? Nobody KNOWS; its a SECRET. Stay tuned for more SURPRISES.
 
#57
Tekky said:
The controller has an IR filter over the front of it so that it only sees IR light. Wouldn't make a very good camera.

Plus, it looks like it has some image processing built in so that it's only sending information about the points (LEDs) it sees; it's not sending back a whole image to the console; that would require too much communication (be slow and drain battery power).

With 2 LEDs in the sensor bar, the controller can tell what direction it is pointed with respect to the LEDs, and it can tell how far away it is from the bar (if it can see both LEDs). It cannot detect its own position in 3D space, though. You'd need a minimum of 3 non-collinear LEDs for that purpose (4 is better).
Ok so the optical sensor can nail 4 degrees of freedom, limiting it's possible position and orientation to being on a sphere (think about it....hint; the centre of the sphere is not the centre of the sensor bar) with the orientation being different at each point on the sphere. Therefore by using the tilt sensors (2 degrees of freedom) it can calculate where on the sphere it lies, double checking with the accelerometers (another 4 degrees of freedom). Ingenius, I wouldn't have thought of this and I'm very impressed Nintendo has got this pointing device working as well as it seems to from the demos.

edit; It actually only tells it's exact position and orientation within a reflection...although that'd be through the other side of the TV. This problem can also be solved by making each end of the sensor bar "look" different, e.g. different wavelength.

Unfortunately this means that the remote can't accurately keep track of its location without being pointed at the sensor bar.
 

Gahiggidy

My aunt & uncle run a Mom & Pop store, "The Gamecube Hut", and sold 80k WiiU within minutes of opening.
#58
So, when are developers gonna start leakign details of how this thing works? I'm assuming they have development tools that give read-outs on the coordinates and stuff as this thing is being used. I'd like the know hte range and limits of its spatial measurements.
 
#59
PkunkFury said:
The sensor bar is for locking the position of the remote in 3 space. It provides an external reference so the remote can calculate it's position in relation to the sensor nodes. The accelerometer data can be used to help when the sensor bar fails. if you wave the controller around and point the remote at the sensor again, the game (hopefully) can go off of accelertion data until the tracker bar finds the eye again. The two work together to try and cover all necessary situations with good accuracy.
I have to wonder if maybe the technology is a lot more falible than initially believed. Most of the games that seemed to require tracking got mixed impression, whilst those that used only gestures were generally well recieved.
 
#60
Tekky said:
The controller has an IR filter over the front of it so that it only sees IR light. Wouldn't make a very good camera.

Plus, it looks like it has some image processing built in so that it's only sending information about the points (LEDs) it sees; it's not sending back a whole image to the console; that would require too much communication (be slow and drain battery power).

With 2 LEDs in the sensor bar, the controller can tell what direction it is pointed with respect to the LEDs, and it can tell how far away it is from the bar (if it can see both LEDs). It cannot detect its own position in 3D space, though. You'd need a minimum of 3 non-collinear LEDs for that purpose (4 is better).
It's great to hear more about the sensor. It does sound unlikely that the remote would send full camera images when in motion tracking mode. Do you think it would be possible for a sensor to be developed that switches between a mode that sends tracking data images and a mode that sends photo images?

As far as the IR detecting the position, it is at least detecting the controller's position in 2 of three directions using the LEDs. I know of devices that can be tricked into sensing the third direction as well with a reverse configuration (2 sensors tracking one LED) so maybe it's possible they've found a trick to get all positional imformation from the sensor bar with one sensor and 2 LEDs as well. If they haven't that would explain why we havn't seen a lightgun style game yet, and the IR sensors would only be reinforcing data for two of the positions.

kangu said:
I have to wonder if maybe the technology is a lot more falible than initially believed. Most of the games that seemed to require tracking got mixed impression, whilst those that used only gestures were generally well recieved.
most of the negative impressions I've seen sounded more like user error than anything else. People just weren't used to the tech, didn't understand how to point the wand, and didn't like the sensetivity to the target on the screen. Other errors sounded like software problems and problems with the setup (particularly Red Steel). Videos have shown the wand working very well as a pointer with extreme precision.

The only hardware related errors I've heard people mention invovle the IR tracking "spazzing out". The IR tracking is searching for the two LED nodes whenever you play in pointer mode. If the sensor doesn't find those two LED nodes it has to guess what it should be doing based on whatever IR imformation and inertial data it receives. People who waved their arms around or stood too close to the screen, likely occluded one of the tracking targets, confusing the device. Another reason the device would spaz out would be if the IR sensor detects more than one LED target. Other targets could be more sensor bars or even retro reflective material such as the emblems on gym shoes. A lot of people mentioned spazzing out in relation to Red Steel, which was the game that had multiple sensor bars set up in a small area without curtains in between them. Nintendo supposedly rearranged these boths during the last day when they realised it was causing problems (how they missed this in the first place is beyond me)
 
#61
PkunkFury said:
It's great to hear more about the sensor. It does sound unlikely that the remote would send full camera images when in motion tracking mode. Do you think it would be possible for a sensor to be developed that switches between a mode that sends tracking data images and a mode that sends photo images?

As far as the IR detecting the position, it is at least detecting the controller's position in 2 of three directions using the LEDs. I know of devices that can be tricked into sensing the third direction as well with a reverse configuration (2 sensors tracking one LED) so maybe it's possible they've found a trick to get all positional imformation from the sensor bar with one sensor and 2 LEDs as well. If they haven't that would explain why we havn't seen a lightgun style game yet, and the IR sensors would only be reinforcing data for two of the positions.



most of the negative impressions I've seen sounded more like user error than anything else. People just weren't used to the tech, didn't understand how to point the wand, and didn't like the sensetivity to the target on the screen. Other errors sounded like software problems and problems with the setup (particularly Red Steel). Videos have shown the wand working very well as a pointer with extreme precision.

The only hardware related errors I've heard people mention invovle the IR tracking "spazzing out". The IR tracking is searching for the two LED nodes whenever you play in pointer mode. If the sensor doesn't find those two LED nodes it has to guess what it should be doing based on whatever IR imformation and inertial data it receives. People who waved their arms around or stood too close to the screen, likely occluded one of the tracking targets, confusing the device. Another reason the device would spaz out would be if the IR sensor detects more than one LED target. Other targets could be more sensor bars or even retro reflective material such as the emblems on gym shoes. A lot of people mentioned spazzing out in relation to Red Steel, which was the game that had multiple sensor bars set up in a small area without curtains in between them. Nintendo supposedly rearranged these boths during the last day when they realised it was causing problems (how they missed this in the first place is beyond me)

all this goes to shwo why 1 wiimote can not be an accesory, and 2 nintendo decided to make the wiiless powerful, casue developers will turn out good graphics with shitty control setup
 
#62
Hmm... I don't know about the sensor bar having only two IR leds, wouldn't it increase accuracy if there were more? 2x3 array (six leds) would give pretty good idea of the angle and distance I think. The bar is pretty long after all.
 

catfish

I have a foreskin yet I do not have AIDS
#63
so what happens to these little things when I accidently and inevitably toss my controller across the room?
 
#64
catfish said:
so what happens to these little things when I accidently and inevitably toss my controller across the room?
Being attached to the nunchaku or remote (depending on which piece you're throwing), it wildly swings back toward you, wraps around your neck, and chokes you to death; a fitting reward for your gaming incompetence!
 
#66
SpoonyBard said:
Hmm... I don't know about the sensor bar having only two IR leds, wouldn't it increase accuracy if there were more? 2x3 array (six leds) would give pretty good idea of the angle and distance I think. The bar is pretty long after all.
I think the position of any other sensors would be easy to calculate from the sensor at each end, there would be a slight effect where the central point was to one side, but I think it'd be too small to be of any use. As for having 2 sensors vertically, do you think the bar is tall enough to have any effect?
 
#67
PkunkFury said:
most of the negative impressions I've seen sounded more like user error than anything else. People just weren't used to the tech, didn't understand how to point the wand, and didn't like the sensetivity to the target on the screen. Other errors sounded like software problems and problems with the setup (particularly Red Steel). Videos have shown the wand working very well as a pointer with extreme precision.

The only hardware related errors I've heard people mention invovle the IR tracking "spazzing out". The IR tracking is searching for the two LED nodes whenever you play in pointer mode. If the sensor doesn't find those two LED nodes it has to guess what it should be doing based on whatever IR imformation and inertial data it receives. People who waved their arms around or stood too close to the screen, likely occluded one of the tracking targets, confusing the device. Another reason the device would spaz out would be if the IR sensor detects more than one LED target. Other targets could be more sensor bars or even retro reflective material such as the emblems on gym shoes. A lot of people mentioned spazzing out in relation to Red Steel, which was the game that had multiple sensor bars set up in a small area without curtains in between them. Nintendo supposedly rearranged these boths during the last day when they realised it was causing problems (how they missed this in the first place is beyond me)
I'd say close to 100% of the problems people had with the controller are because at E3, the controllers were not calibrated against the position of the sensor bar relative to the TV. The first thing most people do is point the remote at the screen and expect to have the cursor be at the center of the screen, where the remote is pointing. Unfortunately, since no calibration was performed, the middle of the sensor bar was "dead center" and everything was relative to that, so pointing at the center of the screen was really "due north" to the controller.

I think that most gamers would adjust quickly even without calibration (after about a minute I had no problem playing Red Steel) - but if Nintendo wants to appeal to the mass market, they will definitely have to fix this.

The remaining problems with the controllers were probably because if you pointed the controller up, it would pick up the fluorescent lights in the convention center as the sensor bar and the cursor would spaz out.
 
#68
The Friendly Monster said:
I think the position of any other sensors would be easy to calculate from the sensor at each end, there would be a slight effect where the central point was to one side, but I think it'd be too small to be of any use. As for having 2 sensors vertically, do you think the bar is tall enough to have any effect?
Yeah, the bar is not very tall, but I guess it all depends on the accuracy of the sensor inside the controller. I can't wait for someone to take these things apart and post pictures :D
 
#71
I still dont see how that LED array in the pic (if it's the one that looks like [o oo o o oo o]) could give elevation info. You need points at a different height.

Why do people think the Wiis weren't calibrated right at E3? You'd think on it's first public outing Nintendo would've made sure everything was perfect.
 
#72
yeah, I noticed those 8 LEDs in some of the pictures as well. The tracking chip is a multi trageting chip so I bet the multiple lights are designed for better tracking with timing. Can't wait until someone takes one of these apart
 
#73
Shmmeee said:
Why do people think the Wiis weren't calibrated right at E3? You'd think on it's first public outing Nintendo would've made sure everything was perfect.
Could be in some cases it was just figured that light gun pointing wasn't necessary. Just as we don't actually roll our mice around on a PC monitor, the wiimote could still be used as a cursor without extra calibration if the bar is in a vaguely good place.

There's also the fact that one calibration would not fit all. The calibration process would give it the necessary information about where the screen is relative to the controller... but that itself would differ depending on where a person is standing, how tall they are, and how they hold the controller.
 
#74
Shmmeee said:
I still dont see how that LED array in the pic (if it's the one that looks like [o oo o o oo o]) could give elevation info. You need points at a different height.

Why do people think the Wiis weren't calibrated right at E3? You'd think on it's first public outing Nintendo would've made sure everything was perfect.
Well I can't comment on the exact technology that nintendo is using, but an example of a system with only 2 points of input giving 3d results in hearing. Despite the fact you only have 2 ears you can hear things and locate them in 3 dimensions because of timing and intensity. The Wiimote could probably be doing the same thing except in reverse where it picks up the 2 points and calculates.
 
#75
Shmmeee said:
I still dont see how that LED array in the pic (if it's the one that looks like [o oo o o oo o]) could give elevation info. You need points at a different height.

Why do people think the Wiis weren't calibrated right at E3? You'd think on it's first public outing Nintendo would've made sure everything was perfect.
Well if it really is like a camera that films those LEDs, you can easily detect if the points are on the top or the bottom of the image, no? But why are they symmetrical? how can the Wiimote detect if it isn't flipped 180° when the Wii is powered on? (yeah I know, noone would do that)
 
#76
:Motorbass said:
Well if it really is like a camera that films those LEDs, you can easily detect if the points are on the top or the bottom of the image, no? But why are they symmetrical? how can the Wiimote detect if it isn't flipped 180° when the Wii is powered on? (yeah I know, noone would do that)
The tilt sensors.
 

Galactic Fork

A little fluff between the ears never did any harm...
#80
JoshuaJSlone said:
Could be in some cases it was just figured that light gun pointing wasn't necessary. Just as we don't actually roll our mice around on a PC monitor, the wiimote could still be used as a cursor without extra calibration if the bar is in a vaguely good place.

There's also the fact that one calibration would not fit all. The calibration process would give it the necessary information about where the screen is relative to the controller... but that itself would differ depending on where a person is standing, how tall they are, and how they hold the controller.
This would also go along with each family member having their own Wiimote. The Wii not only turns on with your personal settings of screen color/skins and stuff, but also calibrated to how you usually play in respect to the sensor bar.


EDIT: quoted the wrong post first.
 
#81
Originally Posted by BigGreenMat:
Well I can't comment on the exact technology that nintendo is using, but an example of a system with only 2 points of input giving 3d results in hearing. Despite the fact you only have 2 ears you can hear things and locate them in 3 dimensions because of timing and intensity. The Wiimote could probably be doing the same thing except in reverse where it picks up the 2 points and calculates.

This would also go along with each family member having their own Wiimote. The Wii not only turns on with your personal settings of screen color/skins and stuff, but also calibrated to how you usually play in respect to the sensor bar.
Woah slow down. Hearing is so much more complicated than that. Your ears are directional, given that they point in a certain direction, your brain can calculated that a slightly muffled sound is more likely to be coming from behind you, also your body picks up vibrations which help you ascertain a lot. Finally your brain does a LOT of guesswork, usually it is right given that our natural experiences are limited. However its easy to trick your ears into "hearing things". Don't think that you can use a 4 dimensional input to get a 6 dimensional output. It's just not true.
 
#82
Here's my technical question :p ... How long is the sensor bar? Some shots make it look quite long to the point that they'd have to have an elongated (cardboard) box to fit the whole lot (console+controller etc) in.

Now, a relatively long cardboard box when Wii is only 3 dvds big? That there is gonna be food for thought come November. Indeed.
 
#83
:Motorbass said:
Well if it really is like a camera that films those LEDs, you can easily detect if the points are on the top or the bottom of the image, no? But why are they symmetrical? how can the Wiimote detect if it isn't flipped 180° when the Wii is powered on? (yeah I know, noone would do that)
Good point, tilt sensors could work out if it's pointing down or up towards the bar as well. THey could also deal with the 180° flip I assume.

Well I can't comment on the exact technology that nintendo is using, but an example of a system with only 2 points of input giving 3d results in hearing. Despite the fact you only have 2 ears you can hear things and locate them in 3 dimensions because of timing and intensity. The Wiimote could probably be doing the same thing except in reverse where it picks up the 2 points and calculates.
I actually work with sound location software, you need a difference in height to ascertain exactly where a sound is coming from, usually by tilting your head. I'm fairly sure it's done how the first post I quoted says. It's a really neat system.

You couldn't do it with just two points as you wouldn't be able to tell if you were moving further away or to the side from the distance between lights, that's why you need the 2x4 arrangement you see in that picture.

EDIT:
Here's my technical question :p ... How long is the sensor bar? Some shots make it look quite long to the point that they'd have to have an elongated (cardboard) box to fit the whole lot (console+controller etc) in.

Now, a relatively long cardboard box when Wii is only 3 dvds big? That there is gonna be food for thought come November. Indeed.
It could be broken down into two or three parts you put together.