Support NeoGAF

[gofreak]

so wouldnt that make it less precise when aiming?not talking about the other capabilities. I've read that the wiimote had an infrared camera with high resolution. Or did i read wrong?... i think it was that oriental guy who made those awsome demos using the wiimote...he said something about the wiimote having a hi-res(1024x780 or something) infrared camera

I think people mightn't be happy comparing the two here, but the native resolution I believe is 128x96 on the wiimote camera. The image is resized up to 1024x768 for processing.

Of course, because the camera is on the controller, the resolution is all 'dedicated' to what it's orientated at, so it's not really comparable apples-to-apples with the resolution of a camera observing a controller.

In any case, I think the Sony motion controller is perhaps more reliant on colour resolution than on image resolution for this. I think they may use the RGB intensities, compared against a known rgb intensity model of the sphere, to calculate the centre of the sphere at a sub pixel level. They're not looking at the pixels and going 'oh, the centre of the sphere is at this pixel'.

 

[Lonely1]

so wouldnt that make it less precise when aiming?not talking about the other capabilities. I've read that the wiimote had an infrared camera with high resolution. Or did i read wrong?... i think it was that oriental guy who made those awsome demos using the wiimote...he said something about the wiimote having a hi-res(1024x780 or something) infrared camera

No, Arc is perfect and far superior to anything before and after it. If you disagree, you are a wiitard that sleep with Kirby pillows... \sig, Eyepatched Gaf...

 

[BruceWayneIII]

Damn proxy blocking everything, can't see the GIF, someone rehost on flickr or something please

Grab it before I remove it again:

http://farm5.static.flickr.com/4049/4422063477_f1cc46d16b_o.gif

 

[donny2112]

I've read that the wiimote had an infrared camera with high resolution.

It needs to see two dots. How high a resolution does that really require?

 

[UntoldDreams]

so wouldnt that make it less precise when aiming?not talking about the other capabilities. I've read that the wiimote had an infrared camera with high resolution. Or did i read wrong?... i think it was that oriental guy who made those awsome demos using the wiimote...he said something about the wiimote having a hi-res(1024x780 or something) infrared camera

Precision in the PS3 system is not solely based on the camera. The camera provides a point of reference not an exact pixel point in 3D space.

For ARC the camera simply shows where in the room the Ball currently is, providing essentially a Point A to Point B line of sight reference in 3D VECTOR space.

A-------B (any motion done with linear & rotational measurements can pinpoint you)

Throw in a bunch of data from the motion & rotational sensors with a Point to Point reference and you get 1:1 motion which also must account for sensor inaccuracy using the visual data.

 

[jaypah]

No, Arc is perfect and far superior to anything before and after it. If you disagree, you are a wiitard that sleep with Kirby pillows... \sig, Eyepatched Gaf...

:lol really? maybe this isn't the thread for you.

 

[Lonely1]

I think people mightn't be happy comparing the two here, but the native resolution I believe is 128x96 on the wiimote camera. The image is resized up to 1024x768 for processing.

Perhaps that account for the 300hz said by Iwata and the 120Hz reported by Lee.

 

[dogmaan]

I think people mightn't be happy comparing the two here, but the native resolution I believe is 128x96 on the wiimote camera. The image is resized up to 1024x768 for processing.

Of course, because the camera is on the controller, the resolution is all 'dedicated' to what it's orientated at, so it's not really comparable apples-to-apples with the resolution of a camera observing a controller.

In any case, I think the Sony motion controller is perhaps more reliant on colour resolution than on image resolution for this. I think they may use the RGB intensities, compared against a known rgb intensity model of the sphere, to calculate the centre of the sphere at a sub pixel level. They're not looking at the pixels and going 'oh, the centre of the sphere is at this pixel'.

depth should be pretty easy to calculate as well

Radius of Sphere = Distance from Camera

 

[beast786]

:lol really? maybe this isn't the thread for you.

Trust me. He aint going anywhere.

 

[Lonely1]

Precision in the PS3 system is not solely based on the camera. The camera provides a point of reference not an exact pixel point in 3D space.

For ARC the camera simply shows where in the room the Ball currently is, providing essentially a Point A to Point B line of sight reference in 3D VECTOR space.

A-------B (any motion done with linear & rotational measurements can pinpoint you)

Throw in a bunch of data from the motion & rotational sensors with a Point to Point reference and you get 1:1 motion which also must account for sensor inaccuracy using the visual data.

The problem I see is that mechanical tracking isn't as good as optical tracking. So, If arc relies on mechanical tracking for pointing, It would be less precise than Wiimote's pure optical pointer.

 

[jaypah]

Trust me. He aint going anywhere.

i know. but why stick around for something that obviously frustrates you?

 

[Lonely1]

It needs to see two dots. How high a resolution does that really require?

The Wii-mote can track up to 4 moving dots at each hertz.

 

[Lonely1]

i know. but why stick around for something that obviously frustrates you?

Because im very interested in Arc. I will just ignore the sony-loyalist and their ignorant cheapshots that don't bring anything to the conversation from now on.

 

[electroshockwave]

Grab it before I remove it again:

http://farm5.static.flickr.com/4049/4422063477_f1cc46d16b_o.gif

How exactly does it detect the controller spinning like that?

 

[Oozinator]

The Wii-mote can track up to 4 moving dots at each hertz.

"each hertz" ?? Do you mean each second or each frame ?

The hertz (symbol: Hz) is the SI unit of frequency defined as the number of cycles per second of a periodic phenomenon.

 

[Lonely1]

How exactly does it detect the controller spinning like that?

The gyroscopes. Wm+ can track rotation like that, but when I tried it, the attachment disconnected some times. Clrearly isn't designed for such wild movements, unlike Arc it seems .

 

[The Chef]

Grab it before I remove it again:

http://farm5.static.flickr.com/4049/4422063477_f1cc46d16b_o.gif

Wow that's really impressive.
I haven't played a Wii since Metroid but that looks far more accurate.

 

[gofreak]

Perhaps that account for the 300hz said by Iwata and the 120Hz reported by Lee.

I think it was Takeda who talked about a sensor with '200-300 signals' per second. But I think he was talking about the accelerometer? It's 3 axis, samples at 100hz = 300 'signals' a second?

From what I've read everything in the wiimote seemed to be synched to sample and refresh rates of 100hz, including the ir camera. This is probably the best or most comprehensive dissection of it I've seen: http://uxweek.com/slides/wiimote-projects-uxweek.pdf

Also, re. your problem earlier with the m+ spinning and 'disconnecting'...is it possible your spinning was exceeding the rate supported by the gyro? When I was trying to find reference to this '300hz', I came across the Iwata Asks M+ interview, and it's mentioned there that the gyro goes up to 1600 degrees of motion a sec - i.e. '4 and half' full revolutions of the controller per sec. Maybe with your spinning you were going faster than that?

Anyway...verging way OT I guess...

 

[Lonely1]

"each hertz" ?? Do you mean each second or each frame ?

No, the WIi-mote would return the coordinates of up to four points each time it queries to do so by the host.

 

[Baki]

The problem I see is that mechanical tracking isn't as good as optical tracking. So, If arc relies on mechanical tracking for pointing, It would be less precise than Wiimote's pure optical pointer.

ARC has sub pixel accuracy. We saw it at E3.

 

[UntoldDreams]

ARC is rather interesting though when you consider the math involved.

Here are two point in space lines. As you can see the PSEye Camera at point A can tell highly accurately that you are on a vector from point A to B. However, it can only "approximate" the distance from A to B at best using the ARC light bulb size and intensity.


A--------------------B


A
...\
.....\
.......\
.........B

This is a limitation of using a point to point solution. The Wiimote sensorbar has two points of light to help triangulate depth with the caveat that this forces the wiimote to have to SEE the sensor bar all the time if you want to know the distance from A to B
Meaning the wiimote pointing upwards makes the wiimote "blind" for positional calibration.

Sony however, has a secondary solution. Since they already know the 3D vector position is along the line from A--B that really helps pinpoint where you are in 3D space.

Any motion NOT along the A-B line basically can pinpoint you in 3D space.

Example: If you move 2 inches to the right and you are close to the camera we see a large accurate movement of the ball on a pixel level. If you move 2 inches to the right and we see very little movement then we know you are far away from the camera in addition to the ball being smaller.

If you are too far from the camera the movement will be so tiny that ARC will have to rely only on sensor data because we already know that you are "too far" from the camera to provide any useful dynamic calibration data.

All this motion data aggregated together and churned out with a tiny slice of SPU power on the cell processor can provide 1:1 accuracy at high speed.

 

[donny2112]

when I tried it, the attachment disconnected some times.

Was it locked in place with the covering on?

 

[Lonely1]

I think it was Takeda who talked about a sensor with '200-300 signals' per second. But I think he was talking about the accelerometer? It's 3 axis, samples at 100hz = 300 'signals' a second?

From what I've read everything in the wiimote seemed to be synched to sample and refresh rates of 100hz, including the ir camera. This is probably the best or most comprehensive dissection of it I've seen: http://uxweek.com/slides/wiimote-projects-uxweek.pdf

Also, re. your problem earlier with the m+ spinning and 'disconnecting'...is it possible your spinning was exceeding the rate supported by the gyro? When I was trying to find reference to this '300hz', I came across the Iwata Asks M+ interview, and it's mentioned there that the gyro goes up to 1600 degrees of motion a sec - i.e. '4 and half' full revolutions of the controller per sec. Maybe with your spinning you were going faster than that?

Anyway...verging way OT I guess...

"The built-in processor uses 8x subpixel analysis to provide 1024x768 resolution for the tracked points." (Wiibrew.org). It's more than just upscalling the internal image. I think that what you say about wiimote rotational tracking speed limit would cause unprecise tracking, no disconnection.

 

[Lonely1]

Was it locked in place with the covering on?

Yes, but the connection isn't tight by design.

 

[SecretBonusPoint]

Yeah okay, that sword-spin has my complete attention btw. Thats some seriously good looking tracking in place. Sony needs to start the public hype train from now. Get Jonathon Ross in to leak some shit (from his pants!).

 

[jaypah]

Yeah okay, that sword-spin has my complete attention btw. Thats some seriously good looking tracking in place. Sony needs to start the public hype train from now. Get Jonathon Ross in to leak some shit (from his pants!).

the crazy thing is they've shown equally impressive stuff before. i recall them doing a similar demo and turning the wand into a stop sign, tennis racket and a mace. it all seemed really fluid. it's like they're just sitting on this stuff when they should be shouting it from the rooftops. if there's going to be a motion control showdown and you have something that good the people need to know. i know, i know. all in due time.

 

[Sidewinder]

Wouldn't it be cool to have a dueling mini-game in Red Dead Redemption using the ARC and flipping your controller on the palm of your hand? I see many broken ARCs in the future because of guys throwing and flipping it all over the place.

 

[Sidewinder]

the crazy thing is they've shown equally impressive stuff before. i recall them doing a similar demo and turning the wand into a stop sign, tennis racket and a mace. it all seemed really fluid. it's like they're just sitting on this stuff when they should be shouting it from the rooftops. if there's going to be a motion control showdown and you have something that good the people need to know. i know, i know. all in due time.

Yes exactly, I think they already showed this stuff during the E3 reveal?

 

[SecretBonusPoint]

Yeah but that was all simple stuff. That sword spinning is like "holy shit it may as well be real" 1:1 stuff.

But yeah, Sony seems to be coy with presenting all this stuff as commercial ready demo reels but hopefully thats about to change. I remember being massively impressed with the handwriting precision from E3 (after being shown Natals ELEPHANT SPLASH shit).

 

[Lonely1]

the crazy thing is they've shown equally impressive stuff before. i recall them doing a similar demo and turning the wand into a stop sign, tennis racket and a mace. it all seemed really fluid. it's like they're just sitting on this stuff when they should be shouting it from the rooftops. if there's going to be a motion control showdown and you have something that good the people need to know. i know, i know. all in due time.

The fact that Natal has more positive buzz going for it blows my mind...

 

[BritBloke916]

The problem I see is that mechanical tracking isn't as good as optical tracking. So, If arc relies on mechanical tracking for pointing, It would be less precise than Wiimote's pure optical pointer.

I completely understand where you're coming from. Arc certainly allows "sub-pixel" accuracy with regards to X/Y/Z position, but the orientation of the wand may only be determinable via the gyros. I seem to remember reading a discussion on GAF some time ago in which the detection of wand orientation was discussed, and someone claimed that the very top point of the sphere (in line with the wand itself) was a "bright spot" that could be optically differentiated against the rest of the sphere. If true, this would allow for optical detection of orientation via the camera (which, when augmented with the gyro data, could be very precise). By all accounts all our questions will very soon be answered, so we'll just have to wait and see...

 

[jaypah]

Yeah but that was all simple stuff. That sword spinning is like "holy shit it may as well be real" 1:1 stuff.

But yeah, Sony seems to be coy with presenting all this stuff as commercial ready demo reels but hopefully thats about to change. I remember being massively impressed with the handwriting precision from E3 (after being shown Natals ELEPHANT SPLASH shit).

natal bashing aside i agree. also, the way that the AR stop sign was turning pretty much showed that it could spin in that manner. i guess the guy just didn't think to showcase it as it's probably not that impressive to them anymore. after all of the positive feedback from the sword spinning gif i'm sure they wish they would have just showed that from the start. they should have just come out at E3 09 and said, "look, you can spin shit around like this. fucking crazy, right? see ya next fall bitches!" and i probably would have given them a standing ovation.

The fact that Natal has more positive buzz going for it blows my mind...

the power of marketing.

 

[electroshockwave]

I don't know if this has been posted already but EA Sports Active 2.0 is coming to the PS3, Wii and iPhone. The PS3 version will allow you to download new workouts from PSN. http://www.destructoid.com/ea-sports-active-2-0-coming-to-ps3-snubs-xbox-360-166450.phtml
http://edge-online.com/news/ea-sports-active-20-skips-xbox-360

 

[gofreak]

"The built-in processor uses 8x subpixel analysis to provide 1024x768 resolution for the tracked points." (Wiibrew.org). It's more than just upscalling the internal image. I think that what you say about wiimote rotational tracking speed limit would cause unprecise tracking, no disconnection.

On the rotation stuff, I'm not sure then what's up :|

On the ' 8x subpixel analysis ' I think basically means that the location of the blobs can be located down to granularity of an eight of a pixel using their image analysis on the processor. The capture resolution (128x96), the data fed into that algorithm, is what would be analogous to the eye's capture resolution (although like I said earlier, that's not even really apples to apples given that the camera's in totally different contexts in each case). With the eye/arc stuff they're also calculating the sphere's centre point to a 'subpixel level' though to what level of granularity in terms of pixels is undisclosed. In terms of real-space granularity they've said the margin of error is down to sub-millimetre levels (though even that doesn't tell us too much...sub-mm could mean a margin of error of 9/10s of a mm or 1/10 of a mm, a little bit of a difference ).

 

[dogmaan]

The fact that Natal has more positive buzz going for it blows my mind...

Microsoft has the brute force market strength to push anything onto the market, and wrap it into a nice family friendly (and publisher Executive/Decision Maker friendly) package, I'm surprised Sony have actually been doing as well as they have, what with the majority of multi platform titles advertising being sponsored by Microsoft (FFXIII for example).

The vast majority of Natal has been smoke and mirrors so far, Albeit a very pretty smoke and mirrors presentation of family funtime and hype, I can't really fault Microsoft for shrewd marketing

 

[InterMoniker]

Because im very interested in Arc. I will just ignore the sony-loyalist and their ignorant cheapshots that don't bring anything to the conversation from now on.

The only loyalist I see in here are the nintendo fans grasping at straws and trying to downplay sony's motion control.

You're complaing of ignorant cheapshots yet you make a comment like this

No, Arc is perfect and far superior to anything before and after it. If you disagree, you are a wiitard that sleep with Kirby pillows... \sig, Eyepatched Gaf...

 

[Chittagong]

The fact that Natal has more positive buzz going for it blows my mind...

not really so surprising. one has

- Polished game-like demo
- Visionary product experience video
- Simple gadget under TV
- Dude with oversized sunglasses

other one has

- Tech demo style grey presentation
- No vision video
- Glowing pink ball
- Geek who almost faints on stage

Which one will the general public get excited about it

 

[Lonely1]

I completely understand where you're coming from. Arc certainly allows "sub-pixel" accuracy with regards to X/Y/Z position, but the orientation of the wand may only be determinable via the gyros. I seem to remember reading a discussion on GAF some time ago in which the detection of wand orientation was discussed, and someone claimed that the very top point of the sphere (in line with the wand itself) was a "bright spot" that could be optically differentiated against the rest of the sphere. If true, this would allow for optical detection of orientation via the camera (which, when augmented with the gyro data, could be very precise). By all accounts all our questions will very soon be answered, so we'll just have to wait and see...

I think It would require two bright spots.

 

[UntoldDreams]

Microsoft has the brute force market strength to push anything onto the market, and wrap it into a nice family friendly (and publisher Executive/Decision Maker friendly) package, I'm surprised Sony have actually been doing as well as they have, what with the majority of multi platform titles advertising being sponsored by Microsoft (FFXIII for example).

The vast majority of Natal has been smoke and mirrors so far, Albeit a very pretty smoke and mirrors presentation of family funtime and hype, I can't really fault Microsoft for shrewd marketing

The thing is... Despite what this thread has been about both Motion+ and ARC are in the same family tree.

If you design a motion based mechanic for ARC you can also port it to Motion+ and vice versa most likely with little effort.

Microsoft however, is off in space all by itself. They better have some strong marketing because its not like people are going to easily port Motion+ and ARC titles over to Natal.

 

[Septimus]

Well it's a good thing we'll get a updated look at this thing in action tonight. Can't wait.

 

[Zoramon089]

The only loyalist I see in here are the nintendo fans grasping at straws and trying to downplay sony's motion control.

You're complaing of ignorant cheapshots yet you make a comment like this

I haven't made any comments like that =(

 

[Haunted]

The only loyalist I see in here are the nintendo fans grasping at straws and trying to downplay sony's motion control.

You're complaing of ignorant cheapshots yet you make a comment like this

How telling.

 

[jaypah]

The only loyalist I see in here are the nintendo fans

nah, it's both. but it's very interesting and entertaining. plus there's some nice tech discussion going on too. this thread is great.

 

[belvedere]

I couldn't give a shit less about either of the 2 new motion technologies a year ago but after re-watching the e3 presentation I can say I'm fairly excited.

I was impressed with the Sony e3 presentation more than Microsoft’s. I can at least envision Sony’s tech and how it will be utilized, I really can’t see myself flailing my limbs around with a Natal-like mechanism though.

Natal just seems so limited without a controller in the mix while at least with Sony's approach, there's much more room for variety.

 

[UntoldDreams]

I think It would require two bright spots.

I don't know how accurate or inaccurate the gyroscopes inside Motion+ and ARC are though.

Do they get out of tolerance easily? I certainly know that accelerometers (motion) are not so accurate over time and need visual cameras to recalibrate frequently.

Gyroscopic sensors I have some limited information.

The SIXAXIS and Dualshock3 both have a single rotational gyroscope to play racing games.

That gyroscope is deadly accurate and does NOT get out of calibration somehow.

Whether it simply is highly accurate or they installed a gravity & compass based rotational calibration solution I do not know. I can only say I played racing games for hours and did not have a problem with the controller being out of synch with the steering wheel on TV.

 

[Kaako]

I wanna seem them go apeshit crazy on stage with Arc. Show everything it can do and then some. Stop holding back you bastardos!

 

[Haunted]

wow, I just caught up to the "Avatar looks like a children's pop-up book" comment on the last page.

smh


Also, I see no problem with the name Arc in Germanic Languages.

 

[Audioboxer87]

Can't wait for the conference tonight, it's at 1AM GMT time and I'm up really early tomorrow, but I want to see it

 

[Pristine_Condition]

The only loyalist I see in here are the nintendo fans grasping at straws and trying to downplay sony's motion control.

You're complaing of ignorant cheapshots yet you make a comment like this

I haven't made any comments like that =(

Weren't you one of the guys in the other thread who was claiming that there was no difference in accuracy or lag between Sony's system and Wii Motion+?

...because it seems like that would be trying to downplay this tech.

 

[Blablurn]

Can't wait for the conference tonight, it's at 1AM GMT time and I'm up really early tomorrow, but I want to see it

gonna watch it as well. can't wait.