Old Discussion Thread for all 3 motion controllers

[Naboomagnoli]

The Wii definitely isn't calibrated to the screen. It is calibrated to the emitter("sensor" bar). By that you could say it knows roughly where the screen it, but it knows nothing of the screen dimensions. You can't actually point the remote at a specific spot on the screen and expect the cursor to be placed accordingly. It is all based on position relative to the sensor bar.

My assumption, which I'm very confident of, is that the PS3 motion controller will be essentially exactly the same, except everything will be calibrated to the camera. Maybe it will ask you how big your screen is, what the aspect ratio is, how many inches the camera is positioned in front, behind, above and below the tv, but I doubt it.

Would that be so difficult though? It would only have to be a one-off calibration in the XMB and that's it, every game using the PSmotes would be calibrated. It could even take a picture of the room at that time and check whether the edges have moved, meaning the calibration may have changed and it could prompt you to recalibrate or carry on.

Maybe we should just move back to CRT's?

As long as Natal doesn't have the resolution to deal with finger movements, I just can't see it performing the same tasks as the WMP or PSmote without using some sort of additional control mechanism. The writing demo in the PS3 demo was not only of smaller wrist/hand movements than you'd expect from waving your wrist around in Natal, and with instantaneous feedback, but (crucially) with the ability to quickly and easily take the 'pen' off the 'page' between letters.

 

[betan]

The Wii definitely isn't calibrated to the screen.

It is when the game wants to have better knowledge of screen.
Sometimes calibration is done in the background silently making assumptions on where character is pointing at. (For example a tutorial mission showing enemies one at a time).

In any case wii mote (the original) wasn't accurate enough apparently.

It is calibrated to the emitter("sensor" bar). By that you could say it knows roughly where the screen it, but it knows nothing of the screen dimensions. You can't actually point the remote at a specific spot on the screen and expect the cursor to be placed accordingly. It is all based on position relative to the sensor bar.
My assumption, which I'm very confident of, is that the PS3 motion controller will be essentially exactly the same, except everything will be calibrated to the camera.

My assumption is you are way wrong, as calibration, whether explicit or implicit, is very easy to do if you have accurate 3d readings.

 

[bkilian]

Do we have anywhere a report on what exactly Natal can distinguish/track? They superimposed the skeleton in the demo. Can it follow feet etc.?

Also, it just occured t me, if you can determine a player from their background in an image, you could map a 3D skeleton to it as we know the human proportions, and using inverse kinematics along with some 3D mapping (PS3motes) should be able to derive a good pose. The chief problem there is probably background extraction!

I believe it actually does track skeletal movements. That's how they can continue to track someone even if they're partially occluded.
http://www.engadget.com/2009/06/03/project-natal-video-hands-on-impressions-and-further-details/

 

[ShadowRunner]

What I'm seeing is exactly the same thing the Wii does. He is moving a cursor on the screen by moving the controller. It has nothing to do with pointing at exactly at the point on the screen he wants to shoot. In fact, it doesn't look like he's pointing. I'm not complaining about that. I'm just explaining how it works. It should work great for shooters. It worked very nicely in Medal of Honor: Heroes 2 for the Wii and that was with the vanilla controller, not even motion plus.

I'm just saying, you should be able to do pointing with Natal using your hand since it isn't necessary to have to point at exact spots on the screen. All of the other solutions do it the same way with motion relative to the camera. Essentially you are moving a cursor on the screen in 2D space, and I can't think of a reason you wouldn't be able to do that.

There is a difference, this isnt just using the camera for positioning but also the 1:1 tracking the gyroscopes provide. Take a look at the flashlight demo, i cant see that being possible on either natal or wii.

 

[patsu]

I believe it actually does track skeletal movements. That's how they can continue to track someone even if they're partially occluded.
http://www.engadget.com/2009/06/03/project-natal-video-hands-on-impressions-and-further-details/

Could you also approximate the skeletal view in 2D since human structure is rather standard and rigid ? The question is whether you need to be very accurate for full body tracking.

The other focus would be finger/pointer level tracking for fine grain interaction.

 

[Shifty Geezer]

What I'm seeing is exactly the same thing the Wii does. He is moving a cursor on the screen by moving the controller. It has nothing to do with pointing at exactly at the point on the screen he wants to shoot.

Not true. Wii already has 'point where you want to activate on the screen', present the moment you switch it on and use the interface with the pointer.

If you trace a line along the axis of the Wiimote or simlar controller, it intersects the screen at a point which is fairly accurately known. Natal doesn't look to have any suitable means to determine such an axis of direction for objects pointed mostly towards the screen.

Note this is different to air-writing. Air-writing, as you do with a sparkler firework (maybe only UK members undertand what that is?) will work fine by mapping hand position to screen position. This is achieved through XY displacement, but not XY rotation, which Wiimote and PSMC provide.

Sadly both GT and YT are down so I can't check the videos. I can't remember if the PSMC demo was all wrist action or arm action.

 

[Shifty Geezer]

Could you also approximate the skeletal view in 2D since human structure is rather standard and rigid ? The question is whether you need to be very accurate for full body tracking.

It would be to some extent, if you can find the contours of the player. You can predict through IK where forwards of the body a hand would be if the dimensions of the arm were very narrow, for example. Face recognition provides enough clues to follow head motions, as demonstrated by Toshiba with their virtual makeover demo.

 

[Scott_Arm]

Not true. Wii already has 'point where you want to activate on the screen', present the moment you switch it on and use the interface with the pointer.

If you trace a line along the axis of the Wiimote or simlar controller, it intersects the screen at a point which is fairly accurately known. Natal doesn't look to have any suitable means to determine such an axis of direction for objects pointed mostly towards the screen.

I don't understnad what you mean. Which axis on the controller and which point on the screen?

I've almost never been able to point at any part of my screen and have the cursor appear at that point when using the Wii. All I know is that the Wii remote sees the sensor bar, but it does not see your screen at all. All it knows is whether the screen is above or below, but nothing else. Maybe you could calibrate to do accurate screen pointing, but I've never seen it used, and I've never seen it in game.

Note this is different to air-writing. Air-writing, as you do with a sparkler firework (maybe only UK members undertand what that is?) will work fine by mapping hand position to screen position. This is achieved through XY displacement, but not XY rotation, which Wiimote and PSMC provide.

Sadly both GT and YT are down so I can't check the videos. I can't remember if the PSMC demo was all wrist action or arm action.

That's true, there probably isn't any rotation, but as a simple pointer is that even important? I was only originally responding to the claim that Natal couldn't be used as a pointing device. I think roughly you could map any part of your body to XY movement to use as an XY cursor on screen, which is a pointer in the same way that the mouse is a pointer and the same way the Wii remote (in my experience) is a pointer.

 

[goonergaz]

I did think of that after I posted though, there's nothing to stop MS from utilizing the built in remote, or someone could even hold a controller I suppose.

except the fact they are pushing that you don't need controllers anymore.

that would be like releasing a controller that you can use as, say, a tennis racket in a tennis game - saying how great and 'realistic' it is - and then releasing a add-on that fixes it because it's wasn't good enough...

 

[ShadowRunner]

That's true, there probably isn't any rotation, but as a simple pointer is that even important? I was only originally responding to the claim that Natal couldn't be used as a pointing device. I think roughly you could map any part of your body to XY movement to use as an XY cursor on screen, which is a pointer in the same way that the mouse is a pointer and the same way the Wii remote (in my experience) is a pointer.

You dont point the wiimote by moving it on the xy axis though, your understanding of how these pointing devices are working is flawed. You can point in different directions with the wiimote wile keeping its xy position the same, its the angle the device is facing/pointing that makes it a pointer.

 

[Scott_Arm]

You dont point the wiimote by moving it on the xy axis though, your understanding of how these pointing devices are working is flawed. You can point in different directions with the wiimote wile keeping its xy position the same, its the angle the device is facing/pointing that makes it a pointer.

But functionally it's exactly the same thing, even in 3D space.

I can see your point if you wanted to do something like the flashlight demo. You want to hold a flashlight and flash it around by turning your wrist to illuminate different areas in the game world. Natal might be able to do something that could use arm direction, but it would be far less natural and intuitive. I was thinking the question of whether could be used as a pointer was referring to its use in selecting a point on screen by moving a cursor or whatever.

 

[22psi]

Unsure if this has been linked, Natal hands on:

http://kotaku.com/5276789/from-stick-figure-to-mind+blowing-controls

"I drove around a bit of the city using my imaginary steering wheel, weaving between cars with no problem, tearing around sharp corners and deliberately plowing into other vehicles. It felt a bit weird to hold my arms out straight in front of me and pretend to drive, but the results were so lag free, so responsive that it didn't bother me."

 

[betan]

Unsure if this has been linked, Natal hands on:

http://kotaku.com/5276789/from-stick-figure-to-mind+blowing-controls

Impressive software (with a crystal sphere maybe?):

All of that was just the system's input, the software took that input and removed lag to make sure that your motions were fed through the system to the game's controls will little to no slow down.

z-tech is awesome though.

 

[RancidLunchmeat]

Yes, I keep reading comments on here that Natal looked 'laggy' and the response time wasn't good, it was a tech demo that doesn't actually work, etc..

All the 'independent' reports I've read from the media people who had the opportunity to use it at E3 have said it is incredibly responsive, so much so that they were oversteering in Burn Out because they were using motions that were too expressive and the precision was so good.

 

[ShadowRunner]

But functionally it's exactly the same thing, even in 3D space.

Its not the same, with simple xy tracking nadal would not be able to reproduce the flashlight demo at all.

Yes you could control a cursor on screen but it wouldnt be really "pointing", you could also do it with voice controll but that woulnt be pointing either.

 

[Naboomagnoli]

Conversely I read this:

http://news.cnet.com/8301-17938_105-10254684-1.html?tag=mncol;title

Project Natal, the Xbox 360's motion-sensing camera add-on, certainly has potential, but the promo video for it Microsoft showed off was purely a work of science fiction. In it, a happy family enjoys multimedia content, chats with friends, and plays complex interactive games without a controller, just using their bodies. The actual playable demos were a few generations behind that, more reminiscent of the Sony Eye Toy accessory for the PS2--the main example was a simple game where players bat a ball back at the screen by swatting at the air, with just enough lag to be annoying. We're very excited about the potential of this new motion-sensing, face-and-voice-recognizing, camera add-on, but for now the gulf between the reality and prerendered video is sizable.

 

[Scott_Arm]

I'm curious to know the difference of the accuracy in the camera imaging part and the 3D depth sensing. The camera must have fairly good resolution and accuracy to detect differences in facial expression. What's his dude warned people about overcalibrated their smiles/frowns. You should be able to smile or frown normally.

 

[Scott_Arm]

Its not the same, with simple xy tracking nadal would not be able to reproduce the flashlight demo at all.

Yes you could control a cursor on screen but it wouldnt be really "pointing", you could also do it with voice controll but that woulnt be pointing either.

I get your point about the flashlight. I'd already edited my post above.

Edit:

I was thinking of pointing in the physical world to select a point on screen, like you'd use a laser pointer during a presentation. There are ways to do this using your hand or arm that would functionally be the same. What you're talking about is in-game, and Natal could not do that from what I've read. I'm not sure which obonicus was referring to when he questioned whether it could be used as a "pointer."

 

[Naboomagnoli]

I've always thought that it could be used to point in a crude way, but would be more effective in a situation where you wave your hand in the right position. In that sense it could be used in a menu (such as the dashboard), but not for anything that is time critical or requires anything like the sort of pinpoint accuracy you could glean from the PSMote/WMP system - it just isn't playing to the Natal's strengths at all.

 

[patsu]

Yes, I keep reading comments on here that Natal looked 'laggy' and the response time wasn't good, it was a tech demo that doesn't actually work, etc..

The steering wheel demo should work because it's easy to track the extended fists in front of the players using the zCam.

All the 'independent' reports I've read from the media people who had the opportunity to use it at E3 have said it is incredibly responsive, so much so that they were oversteering in Burn Out because they were using motions that were too expressive and the precision was so good.

As long as the camera is focused on the fists, it should work like a charm (Not sure how multiple player would work; they will have to iron it out). By right, you don't even need to form the skeleton of the players.

The scenarios people debated around here are more for precise control like writing. There are other advanced use cases that remain to be sampled by outsiders here (like tracking Yoga, karate chops, imaging the environment, tracking different types of finger movement).

In general though, if people are oversteering or misjudging the movement, it is unclear whether it's their fault or the controller's without objective measurements.