The non-standard game interfaces discussion thread (move, voice, vitality, etc.)

[Arwin]

Hmm...I dunno, it seems like a lot of that is still far from actual 1:1 motion, even though they claim it to be. If anything, I think that actually illustrates the differences between M+ and Move better than the last video.

It is absolutely 1:1 movement for the bits in the beginning, and where you throw the ball for instance, or where you hit the dummy. However, it also perfectly illustrates that the Wii can only hang on to 1:1 tracking for a few seconds, as all the demoes that show this has the controller position reset itself within a few seconds. The only demoes that don't actually just track rotation.

 

[tha_con]

Where in the sword vs. dummy demo is it not behaving the same as Move?

Maybe we watched a different video, but the sword only mimics his actions, and you can clearly see it lose what it should be doing when he stops his swips. He'll swing left, and it'll continue to go far beyond his hand, because it is still only guessing where it should be. The Wii has no technology to know exactly where the remote is, and therefor can only guess what 1:1 is and needs to constantly recalibrate and guess.

Move has the benefit of the size and position of the orb to know where it is, and how far away it is. It's pretty clear watching that demo the limitations of Wii M+ in comparison to the solution Move is using. It's only true 1:1 for a brief few moments, if that.

Another give away is when he's just moving the sword for a long time. Each time he moves his arm to a specific direction as far as he can, the sword seems to move less than his arm does, like it's guessing where it should stop. Basically it looks like the accelerometer is saying "it's going this fast in this direction, so move there" and it stops at the max input, while the M+ is telling it which was it should be pointing.

 

[Shifty Geezer]

Maybe we watched a different video, but the sword only mimics his actions, and you can clearly see it lose what it should be doing when he stops his swips.

You'll have to give a time reference, because most of the time the Wiimote is out of view in the sword demo.

The Wii has no technology to know exactly where the remote is, and therefor can only guess what 1:1 is and needs to constantly recalibrate and guess.

It has the same technology in principal as Move, except the visual aspect is more limited as it's Wiimote-based and not system-based. When the wiimote is facing the screen, it has the same positional information available as Move, an optical distance measure andgyroscopic orientation measures.

Move has the benefit of the size and position of the orb...

And Wiimote has the distance between, and position of, the IR sources.

Another give away is when he's just moving the sword for a long time. Each time he moves his arm to a specific direction as far as he can, the sword seems to move less than his arm does.

That's a matter of calibration and scale, mapping the real-world space to the in-game space.

 

[Arwin]

It has the same technology in principal as Move, except the visual aspect is more limited as it's Wiimote-based and not system-based. When the wiimote is facing the screen, it has the same positional information available as Move, an optical distance measure andgyroscopic orientation measures.

And Wiimote has the distance between, and position of, the IR sources.

Yes, but a very limited range. It will lose sight of the IR sources really easily - most sword slashes for instance will only have the wii-mote see the IR source for a fraction of a second, and its resolution of seeing these dots is also rather small.

 

[Shifty Geezer]

Yes, I don't refute that. I was refuting tha_con's point that Wii lacks the tech. It lacks the tech to the same degree as Move, but the tech is the same. Move just has better operating parameters with a great FOV for the device tracking. It'll be interesting to compare the two regards dead-reckoning on a completely occluded controller - does the Move's choice of gyro's have an inherent better error margin, or are they working with the same degree of error, and it's just the higher visibility of the orb making the difference?

 

[Arwin]

Yes, I don't refute that. I was refuting tha_con's point that Wii lacks the tech. It lacks the tech to the same degree as Move, but the tech is the same. Move just has better operating parameters with a great FOV for the device tracking. It'll be interesting to compare the two regards dead-reckoning on a completely occluded controller - does the Move's choice of gyro's have an inherent better error margin, or are they working with the same degree of error, and it's just the higher visibility of the orb making the difference?

Marks shows exactly that in the recent Digital Foundry video on Eurogamer. First he creates a smaller circle with his fingers around the light ball which you see being interpreted immediately as the camera perceiving it to be further away. Then he obscures the Move controller and shakes it violently for a while, and you see the sword drift upwards. It stops drifting immediately once he stops shaking though and it's not nearly as much as I expected, so I think even in that regard it's a little more precise than the Wii, but it was a neat demo. Based on this video though I would suggest that the orb is making the biggest difference. The camera has such a higher resolution and the Orb is so much bigger than the small lights in the sensor bar, and the image recognition software can even know the exact size of the Orb, so should basically be able to exactly tell how far the ball is from the camera as well as be able to determine its position really accurately thanks to it being so easy to find the exact center spot of the circle (combined with everything else of course).

 

[tha_con]

You'll have to give a time reference, because most of the time the Wiimote is out of view in the sword demo.

It has the same technology in principal as Move, except the visual aspect is more limited as it's Wiimote-based and not system-based. When the wiimote is facing the screen, it has the same positional information available as Move, an optical distance measure andgyroscopic orientation measures.

And Wiimote has the distance between, and position of, the IR sources.

That's a matter of calibration and scale, mapping the real-world space to the in-game space.

If you want a time reference, look at the 2:15ish mark, when he first hits the dummy. The sword drifts CONSIDERABLY off screen after his swing, and in no way mimics what he's done, just that he's made a swing to the right, and keeps the orientation right. Not 1:1. Maybe a better example comes later when he raises his sword, hits the dummy on the head, and the sword literally goes flying with the dummy.


The Wiimotes sensing of the distance and position of the IR absolutely requires it to be pointing at the sensors at all times, which means you are extremely limited in what you can and cannot do. The two technologies are not the same in principal. You cannot have completely accurate 1:1 movement at all angles to match everything, including distance, unless it is pointed directly at the screen.

Again, Wii M+ only helps with orientation, not travel distance (position). All of those demo's likely use accelerometers to move the sword on screen to it's stopping point, but it's only saying "moving in X direction at Y speed" and calculates the time to move it to the stopping point. That's not 1:1. It's guesstimation, if anything.

Unless the remote has a constant point of reference (the IR bar) then it cannot know it's position, only it's orientation. The technology is limited, and I don't consider that vague similarity to Move to be "the same in principal". Not even close. It's precisely why an extremely limited number of games on Wii use the IR sensors to determine the distance, etc. Because it's simply not a viable control method.

 

[AzBat]

Unless the remote has a constant point of reference (the IR bar) then it cannot know it's position, only it's orientation. The technology is limited, and I don't consider that vague similarity to Move to be "the same in principal". Not even close. It's precisely why an extremely limited number of games on Wii use the IR sensors to determine the distance, etc. Because it's simply not a viable control method.

Couldn't this problem be mitigated somewhat in Wii 2.0 with additional IR bars? Say for instance an IR bar to the left and right of the user?

Tommy McClain

 

[Shifty Geezer]

The camera has such a higher resolution and the Orb is so much bigger than the small lights in the sensor bar, and the image recognition software can even know the exact size of the Orb, so should basically be able to exactly tell how far the ball is from the camera as well as be able to determine its position really accurately thanks to it being so easy to find the exact center spot of the circle (combined with everything else of course).

The two lights of Wii are actually a more accurate solution (when visible), if only the camera wasn't such low quality. They provide a wider point of reference, equivalent to a Move orb of the diameter of the sensor distance (about 16cm between IR sources by my measuring). Given a 640x480 camera, they would give a more accurate distance reading than the sphere. That doesn't solve all the over shortcomings of Wii's design but at least in that respect the advantage can't be given to Move!

If you want a time reference, look at the 2:15ish mark, when he first hits the dummy.

In that demo the sword is held in place, and is only responding to the Wiimote's rotation. You don't know if he twisted the Wiimote around towards himself, but given how the on-screen interaction through the rest of the video works, there's no reason to think it lost it completely there.

The Wiimotes sensing of the distance and position of the IR absolutely requires it to be pointing at the sensors at all times, which means you are extremely limited in what you can and cannot do. The two technologies are not the same in principal. You cannot have completely accurate 1:1 movement at all angles to match everything, including distance, unless it is pointed directly at the screen.

If you are taking 1:1 to mean a perfect match, then no system provides that. Move doesn't. It's a lot of approximations and on-the-fly calibrations, with a margin of error. Wii's margin of error is larger, but it's the same technology in principal, only Sony have executed it more effectively but shifting the optical tracking to the base unit.

Taking your drift explained above, Move's sphere wouldn't come into play at all. The sphere is for positioning, not rotational information. That comes from the gyros. If you have a Move demo where the sword is rotating about its hilt and not being translated, it's the same technology, three gyros, telling the consoles which way the controller is pointing. In fact, Wii has the added information of visual orientation information. If you point Move at the screen and twist it, only the gyro's tell the PS3 its orientation. On Wii you get the gyro info and the triangulation info of the two IR spots which can be used to determine rotation about at least two axis.

Again, Wii M+ only helps with orientation, not travel distance (position). All of those demo's likely use accelerometers to move the sword on screen to it's stopping point, but it's only saying "moving in X direction at Y speed" and calculates the time to move it to the stopping point.

That's true.

That's not 1:1. It's guesstimation, if anything.

1:1 doesn't mean an exact sampling, but non-gesture based tracking. There are margins for error. The lower the margin of error, the better, but as long as the margin for error doesn't destroy the intentions of the gamer, it counts as one ot one. Putting it another way, if Move wasn't being released and we didn't have tech demos, you'd look at this Motion+ demo and be calling it the 1:1 gaming we wanted from Wii in the first place. 1:1 means when we swing the sword right to left, slightly downwards, slightly upwards, it performs those subtly different moves instead of always doing the canned right-to-left sword move.

Unless the remote has a constant point of reference (the IR bar) then it cannot know it's position, only it's orientation. The technology is limited, and I don't consider that vague similarity to Move to be "the same in principal". Not even close.

The difference is the same between a front-engined car and a rear-engined car. The handling is different, but the mechanics are the same. Gyroscopic orientation info; accelerometer movement info; and visual distance and placement info. The only real difference is Wii has a small window of visual calibration whereas Move's is huge, and Wii has additional visual triangulation when pointing at the screen where Move doesn't.

 

[Arwin]

The two lights of Wii are actually a more accurate solution (when visible), if only the camera wasn't such low quality. They provide a wider point of reference, equivalent to a Move orb of the diameter of the sensor distance (about 16cm between IR sources by my measuring). Given a 640x480 camera, they would give a more accurate distance reading than the sphere. That doesn't solve all the over shortcomings of Wii's design but at least in that respect the advantage can't be given to Move!

But it's the fact that it works in almost any orientation and is detected with a far better resolution camera that does give it the advantage. The advantage for the Wii, when limited to the small field of view of the pointer's camera, is very theoretical, and I'm not even certain of the theory. Wouldn't being able to determine the size of the fairly large ball be much easier to determine its position consistently and linearly compared to what you need to do when the wii-mote is, say, held sideways and pointing a little to the left of the two leds and then moved sideways, if we're purely interested in the distance from the screen? (z info) It in fact I have found no evidence that this kind of calculation is even used for determining the wii-mote's Z position, only for pointing.

Also, while you're correct that the PS Eye can't tell the pointing location of the Move, you may be overlooking that unlike the WiiMote, the Move controller has a magnetometer just for the purpose of filling the holes in the gyroscope. In fact, I overlooked it myself when thinking about the relative drift in both controllers - I think the magnetometer is what helps greatly reduce drift when the controller is invisible to the eye. I was actually wondering why the demonstration Marks gave resulted in comparatively little drift when you looked at how much he moved the controller, but forgot about this.

In that demo the sword is held in place, and is only responding to the Wiimote's rotation. You don't know if he twisted the Wiimote around towards himself, but given how the on-screen interaction through the rest of the video works, there's no reason to think it lost it completely there.

Except that the guy who demonstrates it was quoted on this subject elsewhere, quoting up to 40cm of drift in mere seconds if I remember correctly (though I may well not). If I come across the link again I'll post it here.

If you are taking 1:1 to mean a perfect match, then no system provides that. Move doesn't. It's a lot of approximations and on-the-fly calibrations, with a margin of error. Wii's margin of error is larger, but it's the same technology in principal, only Sony have executed it more effectively but shifting the optical tracking to the base unit.

Looking at the Move controller, especially for augmented reality demonstrations, you could have fooled me into believing it is close to a perfect match though!

 

[Shifty Geezer]

But it's the fact that it works in almost any orientation and is detected with a far better resolution camera that does give it the advantage.

Hang on, I wasn't disputing any of that. Move is far and away the better solution. I was only being pedantic regards your comment, "the orb is so much bigger than the little lights." It's not the size of the lght that matters, but the distance between sample points of your known object. In Move's case, it's a few centimetres for the diameter of the ball, and in Wii's case it's 16cms. Everything else favours Move!

...the wii-mote is, say, held sideways and pointing a little to the left of the two leds and then moved sideways, if we're purely interested in the distance from the screen?

Yes, the LEDs alone can't give that info, but coupled with Motion+ to provide orientation, distance can be triangulated.

Also, while you're correct that the PS Eye can't tell the pointing location of the Move, you may be overlooking that unlike the WiiMote, the Move controller has a magnetometer just for the purpose of filling the holes in the gyroscope. In fact, I overlooked it myself when thinking about the relative drift in both controllers - I think the magnetometer is what helps greatly reduce drift when the controller is invisible to the eye. I was actually wondering why the demonstration Marks gave resulted in comparatively little drift when you looked at how much he moved the controller, but forgot about this.

Drift in that case should be a result of accelerometer error, not gyros, and the magentic calibration only applies to orientation. It's a nice idea and I'm sure it helps plenty in some cases.

Except that the guy who demonstrates it was quoted on this subject elsewhere, quoting up to 40cm of drift in mere seconds if I remember correctly (though I may well not). If I come across the link again I'll post it here.

?? How does that change the demo? the hilt is locked in place, and thus is only responding to Wiimote orientation. That the accelerometers may be placing the Wiimote an extra 40cm left or right doesn't have any effect on that.

Looking at the Move controller, especially for augmented reality demonstrations, you could have fooled me into believing it is close to a perfect match though!

It's within the margin of error. Sub-mm accuracy. It's not accurate to within a nanometre or less - it's not knowing exactly where you are, but whereabouts you are. Motion+'s mapping is maybe accurate to withing 5cm instead of 1 mm (and it's probably much better than that), but it's still satisfying the notion of 1:1 motion.

 

[tha_con]

Couldn't this problem be mitigated somewhat in Wii 2.0 with additional IR bars? Say for instance an IR bar to the left and right of the user?

Tommy McClain

Not really. How would the Wii remote be able to tell the difference between the two different sensors? There really would be no applicable solution using the Wii's current technology. It would have to be new tech.

The two lights of Wii are actually a more accurate solution (when visible), if only the camera wasn't such low quality. They provide a wider point of reference, equivalent to a Move orb of the diameter of the sensor distance (about 16cm between IR sources by my measuring). Given a 640x480 camera, they would give a more accurate distance reading than the sphere. That doesn't solve all the over shortcomings of Wii's design but at least in that respect the advantage can't be given to Move!

I don't understand how you can say "if it had this, it would be more accurate". That doesn't make sense. Accuracy doesn't matter if it hinders the way you play / use the device. It doesn't matter in any application if the Wii's solution (which hasn't even been proven to work to any great degree) is better if it can only function while the remote is pointed directly at the screen. It severely hinders what you can and cannot do with it.

In that demo the sword is held in place, and is only responding to the Wiimote's rotation. You don't know if he twisted the Wiimote around towards himself, but given how the on-screen interaction through the rest of the video works, there's no reason to think it lost it completely there.

You must be watching the wrong portion of the demo? The Dummy comes on screen, and is clearly moving around, responding to rotation and acceleration. It is literally all over the place because it has no point of reference for where it should be. It just knows how fast it was moving, what direction, and the orientation of the remote. No position data what so ever. There is plenty of reason to think it lost it completely, because it did.

If you are taking 1:1 to mean a perfect match, then no system provides that. Move doesn't. It's a lot of approximations and on-the-fly calibrations, with a margin of error. Wii's margin of error is larger, but it's the same technology in principal, only Sony have executed it more effectively but shifting the optical tracking to the base unit.

Move provides an extremely close approximation. The solution in the Wii Remote is limited to where the LEDs are in the cameras FOV. In fact, I wouldn't be surprised if there was little to no libraries from nintendo in the SDK that use those LEDs to approximate distance or position in any way. They'd probably be custom from the dev studio.

Taking your drift explained above, Move's sphere wouldn't come into play at all. The sphere is for positioning, not rotational information. That comes from the gyros. If you have a Move demo where the sword is rotating about its hilt and not being translated, it's the same technology, three gyros, telling the consoles which way the controller is pointing. In fact, Wii has the added information of visual orientation information. If you point Move at the screen and twist it, only the gyro's tell the PS3 its orientation. On Wii you get the gyro info and the triangulation info of the two IR spots which can be used to determine rotation about at least two axis.

The "drift" I'm talking about comes after the swing and has nothing to do with rotational info. It is because the Wii Remote, even with Wii Motion+, has absolutely no way to determine position data for any complex movements. It is ALL based on accelerometers and gryos. "I'm going this fast, I'm tilted in this direction". Since this is the only info it has, there is no point of reference for where it really is, and it starts to drift away, at which point he resets the demo. It does this each and every time.

The difference is the same between a front-engined car and a rear-engined car. The handling is different, but the mechanics are the same. Gyroscopic orientation info; accelerometer movement info; and visual distance and placement info. The only real difference is Wii has a small window of visual calibration whereas Move's is huge, and Wii has additional visual triangulation when pointing at the screen where Move doesn't.

Eh, I don't think that's a fair comparison. A better comparison would be to say you have one car that can steer, but it can only go in one direction (Wii can only detect position data / pointer data when pointing at the screen). Another car has the ability to go anywhere (Move can always know the position of the orb as long as it's in view, and can point off screen and still know "where" the pointer is).

The magnetometer is a pretty neat solution because you can point off of the screen, and they are able to give you feedback as to where you are pointing via an indicator. With the Wii remote, I don't know if that's possible, since once you lost the LED's, you lose any positional data / pointing data.

 

[patsu]

I think the deal is whether a persistent 1-1 tracking mechanism can simplify motion tracking and make it more consistent. Specifically, is the tracking fluidal like the Sorcery game. Can we layer the actions together smoothly ? Can we do it without gesture recognition lag ?

WiiSports Resort is brilliant by simplifying the tracking to gyro only. And yet it's so "accurate". That may mean the devs can track another measurement for "layered action/gestures". This is very interesting stuff.

 

[AzBat]

The Kinect.me site* is up and running and has videos showing a lot of people playing Kinect games during the Kinect Tour. Just a heads-up, looks like a LOT those people are kids & they're having a blast.

Anyway, people who play the demos on the tour are given a "Guest Pass #" that they can then use to find a video of them playing the games, which they can then share on Facebook. The site also has an interactive map showing where the Kinect Tour will be visiting.

Tommy McClain

*

Spoiler

Requires SilverLight

 

[Danalys]

I think the deal is whether a persistent 1-1 tracking mechanism can simplify motion tracking and make it more consistent. Specifically, is the tracking fluidal like the Sorcery game. Can we layer the actions together smoothly ? Can we do it without gesture recognition lag ?

WiiSports Resort is brilliant by simplifying the tracking to gyro only. And yet it's so "accurate". That may mean the devs can track another measurement for "layered action/gestures". This is very interesting stuff.

Sorcery seems to have an easy time of it in some regards. A spell isn't instant it's expected to come after a movement. In the game this is an animation which the player themself provides.

 

[patsu]

Their E3 demo was butter smooth. The pacing was just right. The potion drinking, mixing of fire + wind spells and the creation of the wall of fire looked responsive too. Every major action seemed to have instant feedback in the game (e.g., shaking potion turned the light on). Everything was effortless and consistent. It'd be amazing if the entire game plays like this.

 

[patsu]

http://www.mcvuk.com/news/40209/Kinect-pre-orders-boom

Despite initial scepticism over its price, Kinect looks set to be a big hit.

Retailers are reporting high pre-order numbers for the device, which will be available in November for £129.99 bundled with Kinect Adventures.

 

[scently]

http://www.mcvuk.com/news/40209/Kinect-pre-orders-boom

Saw this earlier today, well it seems we are paying £130 then.

 

[patsu]

We spoke about using natural interface games (or gaming principles) for education:
http://gamasutra.com/view/news/2968..._Incorporate_Virtual_Video_Game_Education.php

Baltimore County Public Schools plans to incorporate video game development education into its curriculum with the help of serious game applications, in collaboration with Learning Port Strategies.

Also related:
http://gamasutra.com/view/news/2968...h_Lead_To_Vietnam_Online_Game_Limitations.php

I'd say the Chinese government has similar mindset w.r.t. online gaming.

 

[patsu]

http://www.destructoid.com/source-p...ing-to-retailers-on-september-11-180534.phtml

Sources that wish to remain anonymous have passed along details of employment opportunities for representatives to assist in what is referred to as "the second largest product launch in PlayStations' history." The demo program will have trainees working kiosks in retailers like Best Buy and Walmart, assisting potential customers with the usage of Move, showing off features, and answering questions about the product which ships on September 19. The demos -- scheduled for peak retail traffic hours -- are said to start on September 11 and last through 2011, ending in late January.