Old Discussion Thread for all 3 motion controllers

[Nesh]

Hence, this is why Natal is not just hardware but critically also, software, which is where MSFT strength lie.

I dont see that critical acclaim. The only thing it gets so far is marketing hype and expectations about the potential of the hardware.

The software is any app or game that exploits the available hardware. And that can be either from MS or some other company.

Removing Natal's ability to detect depth, all you are left with is simply any camera and a microphone. And that alone wouldnt have made people excited about the potential.

 

[patsu]

I dont see that critical acclaim. The only thing it gets so far is marketing hype and expectations about the potential of the hardware.

Well, we also hear occasional developer praises and skeptisms, which could mean the system works marvelously within some bounds.

The software is any app or game that exploits the available hardware. And that can be either from MS or some other company.

Removing Natal's ability to detect depth, all you are left with is simply any camera and a microphone. And that alone wouldnt have made people excited about the potential.

You may be right earlier on to say that many of Milo's capabilities can be done without depth perception, but MS will find applications to emphasize Natal's unique features (e.g., already mentioned Xbox Live users teaching Milo collectively, recognizing 3D objects).

Even though speech recognition may not be 100% reliable, the fact that every 360 comes with a headset will allow developers to experiment with interesting features. The hardcore part of their presentation may still have controllers like the traditional 360 controller pad, headset, 3D glasses, etc. augmented by natural interfaces.

Afterall, no one has demanded that the traditional pad should be replaced by a wand. Sony is taking a big risk trying to introduce the wand directly for core gaming (See Socom 4). Unless there are concrete and evident improvements, it is very likely to end up like SIXAXIS (See Lair).

The casual part of MS's presentation may rely solely on controller-free gaming. For some games, these people may not need high specifications. The excitement may come from seeing refreshing applications for casuals, especially female gamers, young and old.

Most US consumers may not have seen "old" tricks like the EyePet's sketch recognition, sing-along, or even EyeToy motion games. If MS can get all of them working (even partially), their massive marketing effort is going to grab a bigger mindshare than Sony.

I also think that a lot can be done to simplify game UI today.

I'd say if the 3 vendors can usher more people into the gaming industry, it should be a win-win for all of us.


EDIT:
http://xbox360.ign.com/articles/109/1095582p1.html

We already knew that Fable III would be at the show, but this comes as the first confirmation of sorts that Milo will make another appearance. Milo, a game where you interact with a virtual boy via Project Natal, was unveiled at E3 of 2009 but has not been heard from since.

 

[goonergaz]

Afterall, no one has demanded that the traditional pad should be replaced by a wand. Sony is taking a big risk trying to introduce the wand directly for core gaming (See Socom 4). Unless there are concrete and evident improvements, it is very likely to end up like SIXAXIS (See Lair).

I don't exactly disagree with this PoV - however, it has been pointed out that the library used is very close to the Wii so any games that will also appear on the Wii should have Move support along with PS3 exclusives (as I'm sure Sony will push for games to support it).

SIXAXIS was a different scenario, it was one of a kind on a new machine that wasn't selling too well.

 

[Silent_Buddha]

I dont see that critical acclaim. The only thing it gets so far is marketing hype and expectations about the potential of the hardware.

The software is any app or game that exploits the available hardware. And that can be either from MS or some other company.

Removing Natal's ability to detect depth, all you are left with is simply any camera and a microphone. And that alone wouldnt have made people excited about the potential.

As I've stated before, the hardware is the least interesting thing about Natal. It's what MS is doing after the camera has captured the image and depth data and before the applications that people will use that is far far more interesting.

At least for me. I'd love to get some hands on time with the server farm they are using to develope the algorithms that Natal will be using to interpret, recognize, and differentiate multiple moving people with intermixed and obscured limbs.

I'd love to see how close the probability numbers come to determining skeletal locations for points (in 3 dimensions) which are completely obscured by baggy/shapeless clothing using only data from other unrelated body parts.

I'm willing to bet that much of the twitchiness we see in some of the demos is MS still fine tuning the number weighting it is giving to probable positions of skeletal points in 3 dimensions.

Regards,
SB

 

[Graham]

I dont see that critical acclaim. The only thing it gets so far is marketing hype and expectations about the potential of the hardware.

The software is any app or game that exploits the available hardware. And that can be either from MS or some other company.

Removing Natal's ability to detect depth, all you are left with is simply any camera and a microphone. And that alone wouldnt have made people excited about the potential.

Having a standardized set of speech, directional audio, motion, depth, video and gamepad controls for developers to use is what makes Natal interesting.

True, any developer could have done a lot of this on their own - and to an extent many developers have done individual parts themselves - but they have almost always been fairly mediocre and (likely) cost a lot of time and money to develop. Often requiring additional hardware as well.

The point is, Natal makes these features available for any game to use, with staggeringly lower cost for developers compared to developing it themselves.

You can see parallels with the DS / DSi. Other systems have had similar features as addons, but by having a standard set of complementary features which a developer can choose to use - then we start to see the weird and risky games come along and expand the market. Games that would otherwise be financially infeasible.
Just look at Brain training, it has sold something like 40 million copies!

 

[patsu]

I don't exactly disagree with this PoV - however, it has been pointed out that the library used is very close to the Wii so any games that will also appear on the Wii should have Move support along with PS3 exclusives (as I'm sure Sony will push for games to support it).

Yes, borrowing/stealing control schemes from Wii may help. However on Wii, the gamers have no other choice. On PS3, they can always fallback on the familiar DS3 (if the game allows them to).

SIXAXIS was a different scenario, it was one of a kind on a new machine that wasn't selling too well.

If SIXAXIS motion control helped to improve say... FPS accuracy or response time, it would have a following of supporters regardless of whether it's one of a kind, or whether the machine was selling well. Most people who liked SIXAXIS talked about "non-traditional" mechanisms like Folklore and Flower. SIXAXIS steering is just about the only mainstream scheme since it resembles real life driving.

Having a standardized set of speech, directional audio, motion, depth, video and gamepad controls for developers to use is what makes Natal interesting.

True, any developer could have done a lot of this on their own - and to an extent many developers have done individual parts themselves - but they have almost always been fairly mediocre and (likely) cost a lot of time and money to develop. Often requiring additional hardware as well.

The point is, Natal makes these features available for any game to use, with staggeringly lower cost for developers, compared to developing it themselves.

Yap, once the boundaries are known, the developers can toy with it to optimize the experience.

At least for me. I'd love to get some hands on time with the server farm they are using to develope the algorithms that Natal will be using to interpret, recognize, and differentiate multiple moving people with intermixed and obscured limbs.

Why does it need a server farm ?

 

[Prophecy2k]

I'd guess to compute the myriad of various different spacial orientations of different sizes of human forms in various different positions.... lots and lots of data that will become the basis of what Natal uses to recognise human positions

 

[Rolf N]

Having a standardized set of speech, directional audio, motion, depth, video and gamepad controls for developers to use is what makes Natal interesting.

Speech is only partitioned away into the Natal umbrella to make that more appealing. On a technical level, there's no reason to limit its reach that way. Every Xbox 360 comes with a headset.

 

[patsu]

I'd guess to compute the myriad of various different spacial orientations of different sizes of human forms in various different positions.... lots and lots of data that will become the basis of what Natal uses to recognise human positions

Yeah, I was thinking they may send the raw data to the server side to "compile" and collate, then send the summarized result down to Natal/360 for on-the-fly recognition. This would make good use of the XBL servers.

[Not sure if the heuristics result will converge this early or at all though ]

Speech is only partitioned away into the Natal umbrella to make that more appealing. On a technical level, there's no reason to limit its reach that way. Every Xbox 360 comes with a headset.

I think a directional audio toolkit may be good for pinpointing sound source, limited voice recognition, and perhaps 3D audio (output) ? ^_^

EDIT:
If Milo is going to be @ E3, MS will definitely follow up with something close or more impressive than the last demo (regardless if the headset will work better).

 

[Love_In_Rio]

According to Peter Molyneux, they are trying to make 3D object recognition work (by having users "teach" Natal collectively).

The depth perception may be useful to get a sense of the surrounding (e.g., scanning your living room for known/learned items).

We'll see how well it turns out. Other than the out-of-the-box items, the learning could take a while to be accurate.

But yes, you may be able to do some/most of the object recognition via 2D image analysis (and sound). Proper lighting may be needed for accurate color perception though.

EDIT: I remember some universities/labs were developing a semantic network of general knowledge for a virtual human. Perhaps their databases can be used here. ^_^

Something like a neuronal network already makes this quite well. See as an example:

http://www.youtube.com/watch?v=QfPkHU_36Cs

A little of training and Natal could detect any learned object.

 

[patsu]

I don't know whether it's a little training or not to start. May be the depth info can speed this up.

If I remember correctly, ASIMO and AIBO use one or two normal cameras to implement their object recognition subsystem.

EDIT:
Found an AIBO link + computer cluster for training:
http://staff.science.uva.nl/~fjseins/isis/aibodemo.html

Short movie showing the Aibo Object Recognition GUI, in which the 'learning' phase is performed for one object, while being connected to four cluster computers in Europe and one in Australia.

I remember Microsoft has an extensive cloud computing platform in several countries. e.g., The one in Singapore can be used by developers to get public data such as weather condition.

 

[AzBat]

Every Xbox 360 comes with a headset.

OK, I've seen this statement made by 2 different people, but it don't make it true. Every 360 does NOT come with a headset. In fact, the Core and Arcade SKUs have NEVER came with a headset. Only the Pro/Premium and Elite SKUs have included the wired headsets.

You can go back to your originally scheduled broadcast.

Tommy McClain

 

[Silent_Buddha]

Yeah, I was thinking they may send the raw data to the server side to "compile" and collate, then send the summarized result down to Natal/360 for on-the-fly recognition. This would make good use of the XBL servers.

Not in the plans at least from what I've read. It's about what Prophecy2k said. I linked to an aritcle about it a few months ago either in this thread or one of the other motion control threads. I'm about to head off to work now, so can't look it up to relink it.

It's basically feeding lots and lots (millions and millions) of static images of people in motion from which the cluster "learns" to predict where each skeletal point is (in 3 dimensions) whether it is visible or not.

Then it comes up with a probability matrix of where the skeletal point is, then uses an algorithm to determine which point to use. Early skeletal tracking systems that were used when demo'ing Natal showed something like the 5 or 10 most probable locations in realtime. So when watching the skeletal tracking it would seem to flicker and jerk a bit. The final output would only be the one chosen however.

Regards,
SB

 

[patsu]

OK, I've seen this statement made by 2 different people, but it don't make it true. Every 360 does NOT come with a headset. In fact, the Core and Arcade SKUs have NEVER came with a headset. Only the Pro/Premium and Elite SKUs have included the wired headsets.

I see. I think it still make sense since we were talking about speech recognition for core gaming (which should be mostly Pro/Premium and Elite folks).

Not in the plans at least from what I've read. It's about what Prophecy2k said. I linked to an aritcle about it a few months ago either in this thread or one of the other motion control threads. I'm about to head off to work now, so can't look it up to relink it.

It's basically feeding lots and lots (millions and millions) of static images of people in motion from which the cluster "learns" to predict where each skeletal point is (in 3 dimensions) whether it is visible or not.

Then it comes up with a probability matrix of where the skeletal point is, then uses an algorithm to determine which point to use. Early skeletal tracking systems that were used when demo'ing Natal showed something like the 5 or 10 most probable locations in realtime. So when watching the skeletal tracking it would seem to flicker and jerk a bit. The final output would only be the one chosen however.

That's the skeleton tracking part.

In his Youtube Milo presentation above, Peter Molyneux also talks about 3D object recognition and using the online portion to continue to learn new objects.

These have been done before, but if MS can tidy up the toolkits as well as making the databases available to all game developers, it would be a powerful platform -- like Google vending GIS data, Facebook vending social data, Amazon vending product catalog, etc. In this sense, it would be an integrated network services + easy-to-use natural interface platform for app development. Much higher level and defensible than mere "higher precision motion controller hardware" pitch.

 

[LightHeaven]

Natal couldn't do that.Tracking is far from being acurate enough.

Not enough accurate from this close of the screen? I really doubt it, and i think it would have a response time advantage over that muscle setup (there's quite a delay in that video).

Off course, if you are talking about having an entire 4-people band playing with natal than i agree with you.

 

[whome0]

PS Move Developer Diary: The Shoot
PSEye camera tracks player's head to dodge missiles.

Anton Mikhailov interview

Q: I know the Wii Remote (with MotionPlus) uses MEMS technology, what about the Move controller? I had read an interview saying that the Wand uses integrated circuits instead…
A: Our controller also uses MEMS technology. MEMS is a fancy term for a machine built into an integrated circuit.

Q: I had also heard some discussion that since the EyeToy is only 120 Hz and the Wiimote camera is 12 MHz that PlayStation Move might not be best suited for picking up the quick motions.
A: Cameras aren’t the only things that allow us to do fast motion. The accelerometers and gyros are very important for this as well. On the Dualshock 3, the motion tracking was up to about 300 degrees per second while on Move, it’s around 2500. It’s really hard to move faster than that without hurting yourself.

Not all MEMS tech is made the same. They're just chips, and ours have quite a bit of improvement. This means that we can track fast motions as well as slow ones.

The simplest way to sum this up is that PlayStation Move is what is known in academia as a 6DOF tracker. This means that it tracks 6 "degrees of freedom" 'absolutely' (3 position -XYZ, 3 angle). The Wii is an "inertial tracker," which means it tracks position relatively.

 

[goonergaz]

Yes, borrowing/stealing control schemes from Wii may help. However on Wii, the gamers have no other choice. On PS3, they can always fallback on the familiar DS3 (if the game allows them to).

Sony have the best of both worlds...maybe they should Sign up Hannah Montana to front their campaign!? :| lol

Seriously tho it would be nice to have both options, that way you get to choose what you prefer.

 

[Rangers]

Huh, apparently natal is not going to be called "Wave" but something else. Six letters is all my source will say.

 

[Arwin]

PS Move Developer Diary: The Shoot
PSEye camera tracks player's head to dodge missiles.

Interesting, because I think this title is also going to be 3D. Adds up nicely!

 

[Billy Idol]

Sony have the best of both worlds...maybe they should Sign up Hannah Montana to front their campaign

it is said/embarrassing that I understand this joke