Natal

I think there would only be very specific games that would be fun played like this. Turn-based strategy games, where you could move units with your hands, and puzzle games. Anything else seems like it would suffer. Pretending to hold an invisible steering wheel is lame as hell, unless you're playing a Wonder Woman game. Shooters would suffer because your aim with your index finger is not going to be as good as it is with a controller/mouse. And how would your character walk around?

It is cool to think that eventually, somehow, this would lead to interactive hologram stuff. Other than that, I'll pass. The Wii's mountain of shitty "motion controlled" games has kind of turned me off to the whole thing.
 
We can speculate it is scripted all we want.. And you will not look foolish to say so.. But the dev says its not scripted, Peter says its not scripted and the people testing the demo say its not scripted.. I mean the thing is live at E3 for testing. Its limited to a set amount of responses but hardly scripted.

This guy tested Burnout and Dodgeball at E3.
http://xbox360.ign.com/articles/989/989269p1.html
 
Last edited:
Pasquale said:
I think there would only be very specific games that would be fun played like this. Turn-based strategy games, where you could move units with your hands, and puzzle games. Anything else seems like it would suffer. Pretending to hold an invisible steering wheel is lame as hell, unless you're playing a Wonder Woman game. Shooters would suffer because your aim with your index finger is not going to be as good as it is with a controller/mouse. And how would your character walk around?

It is cool to think that eventually, somehow, this would lead to interactive hologram stuff. Other than that, I'll pass. The Wii's mountain of shitty "motion controlled" games has kind of turned me off to the whole thing.
Click to expand...

I think, Pasquale, you could ideally take any toy gun you have and use it for your gun, which would be fucking sweet.

TheOsyris said:
We can speculate it is scripted all we want.. And you will not look foolish to say so.. But the dev says its not scripted, Peter says its not scripted and the people testing the demo say its not scripted.. I mean the thing is live at E3 for testing. Its limited to a set amount of responses but hardly scripted.

This guy tested Burnout and Dodgeball at E3.
http://xbox360.ign.com/articles/989/989269p1.html
Click to expand...

I don't think anyone truly doubts the capability for the paint or the dodgeball like or even the racing style game, but the thing peter showed was beyond what our technology supposedly can do. Then again, we are making the assumption that this thing he showed was a true AI in a sense and that it was able to truly think and respond to all things accordingly as well as synthesize its voice to say whatever it can. I don't think it's really doing that necessarily but as Firestreak said where something like recognizing the tone and responding at such and only with so many words/phrases/whatnot it can use. It's not true AI but a damn good simulation that might fool anyone.

Then again, the thing Raes said is true.. we don't have the technology for them to synthesis someone's voice that fast and such, would make voice actors useless but who knows.

I'm hopeful though ^^ I want fun interactive games like that!
 
I'm not saying there isn't a version out there that works. I'm just saying last night nobody knew shit... by now the video has been analyzed and people are sure it's scripted. This is based on there being a point in the fishing game where the screen reacts BEFORE her hands. Demonstrated with this picture, found by GAF.

9gx9p3.gif


Also by the fact that it knows her name, meaning the software was set up with that specific demo run through in mind.

This doesn't mean necessarily that the tech isn't as good as shown, it just means... well, it wasn't really shown to us for real so we have to trust them.

And it's now stated that it reacts to the tone of your voice, not really to what you say. Which means that the AI is NO WHERE near what was implied by the demo, but that has nothing at all to do with the Natal interface. I just am saying that it was partly the AI that really did me in, because Molyeux really laid in that talking up his creation to astronomical levels that I thought they finally cracked a new level of AI. It's just sort of a concept to show off the interface.

It's all well and good if you understand all this and examine the demo with this knowledge in hand. We can still captain the enterprise or play kings quest, only when we talk to people we're going to have limited dialogue options or else the're going to be saying, "I'm sorry I don't understand that." That's just an innate part of voice technology that based on today's technology and the technology for the foreseeable future that demo wasn't really admitting. Wen're not really at all further along than those old PC games where'd you type what you want to say to characters and spend 10 minutes trying to come up with the right line.
 
TheOsyris said:
We can speculate it is scripted all we want.. And you will not look foolish to say so.. But the dev says its not scripted, Peter says its not scripted and the people testing the demo say its not scripted.. I mean the thing is live at E3 for testing. Its limited to a set amount of responses but hardly scripted.

This guy tested Burnout and Dodgeball at E3.
http://xbox360.ign.com/articles/989/989269p1.html
Click to expand...

Burnout and Ricochet aren't the ones we're talking about being scripted though. Those are fairly straightforward applications of the system, there's no need to script those. They're a bit clumsy yet, the Burnout application is basically just a waste of time for most things, and Ricochet has lag/accuracy issues still, but just getting those to work isn't inordinately complicated.

The Milo demo that they put on screen, however, was pretty obviously scripted.
 
Yea I believe the motion tech, which is impressive. I'd want to try it out myself in a non-controlled environment before I sign on but I don't doubt it really.

It's the voice stuff and the AI that was implied with the Milo demo that sorta took it out of a normal technology iteration and made it seem like it was a real revolution. It was a nice fantasy and I really wish I would live to see this a reality. I just don't know. I mean research this AI stuff. There are multiple fields out there trying to solve each individual problem and most of them don't even know how to go about starting the process after decades. Like there's creativity, and inuition and deductive reasoning, knowledge, perception, knowledge, planning, learning, social intelligence... ext ext ext... these are all separate independent fields with lots of people working on them and not really sure what to do. A lot of people speculate it's not even possible... because you have to solve all of them to solve any of them...

So the idea that this guy fresh from releasing Fable 2 last year goes in and does this with a team and uses it for a computer game is not rational.

When I first saw that vid that's what I thought was going on. The interface is great but using it to communicate with a real AI is what would be dramatic. I thought, like probably a lot of other people, that "real AI" is just a natural progression of the smarter and smarter scripts we're getting. I didn't realize it's completely different. It's a different direction programming things to react based on other things happening. Self awareness isn't going to just "happen" like in Terminator because the computer system gets more advanced. This shit is really hard to make happen.
 
Last edited:
Systems to recognize vocal emotional inflection and facial emotional expression aren't new either, to be honest, but this would be the first widespread home application if they can get it to work consistently.

Which ain't easy. Detecting a question in English, for example, is fairly easy, but even that occasionally has hitches with the most systems available. Deadpan delivery of anything leaves the system behind. Facial recognition falls under the same sort of difficulties. Gross expressions can be read with some consistency, but most emotional states are simply beyond the ability of the system to detect. The cues are too subtle and too variable based on the geometry of an individual face.
 
Yea my jokes would be no good with that little boy. What I normally do is say fucked up things with a straight face. Once at work I did a variation of that bit the janitor in Scrubs did with the Chinese baby. Tried to see how long I could go on with the story before people realized I was fucking with them. Without the context that joke will just read like I'm talking. This could be great for kids though.
 
http://xbox360.ign.com/articles/991/991348p1.html?_cmpid=ign105

"fellow IGN'er Tal Blevins stepped up to the TV and was quickly scanned by the Natal camera in a matter of seconds. Once scanned, Molyneux asked Tal to try and peek around the world as if it were real (which was the same familiar country setting that had Milo sitting in a tree swing next to a pond).

Sure enough, the onscreen perspective moved in unison with Tal's head turns -- he looked left and then right and then left again, and each time he did, the screen reacted accordingly, giving Tal a better look at the serene locale.

Now it was finally my turn, and luckily I got a chance to experience the best part -- direct speech with Milo. Molyneux asked me to say my name clearly in front of the screen and then take a step back.

I did both and then when Peter asked me to take a step forward and I did, Milo immediately recognized me -- "Oh, Hello Jeremy!" Milo said before hopping out of his swing and walking up to greet me. Next he quipped, "I see you're wearing blue today," in reaction to the blue shirt that I had on.

I could hear the sound of impressed acknowledgments from the people standing behind me right afterwards"

So again i dont care what Gaf says lol.. People are testing it at E3 and its not scripted. It is limited, sure.
 
Last edited:
I guess I'm just having trouble seeing the application without the intelligence. He can observe you and pick up your mood and understand some specific lines. However if you're running into this guy playing Oblivian... he's going to FEEL a lot realer as you talk to him and such... but he's not going to FUNCTION a lot better for the purposes of pushing the plot along and might actually function worse.

The dialogue tree is there because those are your options, they wrote scripts for these people based on those answers. If he can't talk on his own accord, all you are doing talking to him is fishing for what different dialogue options you have without knowing them.

When I first saw the demo, I saw this woman chatting and it seemed like it understood her. I don't doubt the tech demo of.. it seeming like it understands you. For a tech demo, that's fine. For a game, if it doesn't actually understand you, what function is it?
 
I still expect it's fully canned dialogue (which is inherently limiting), and that they're going to have to work through the aforementioned problems with voice and facial recognition before it's done.

Canned dialogue is limiting, incidentally, because it means that if your name is not on the list, Milo won't be able to say it, nor will it be able to adequately respond to anything that's not in its limited dialogue options.

A big question as to how much they can do with the canned dialogue will depend on how sophisticated the system is for putting it together. If they can fit it in as "I see you're wearing (insert color here) today!", then that will increase the amount of space they have available for other dialogue responses, and not end up with a 10-yr old British boy going on a murderous rampage at the corporate headquarters after repeating the same line 1000 times with single-word variations.

The problem, as we all know if we've ever dealt with an automated phone system, is that there's usually an audible hitch when they do that. Now, enough time in the studio could probably force the poor voice actor to get each line seamless within the context of a particular sentence, but then he's got to do it for every sentence, every emotional inflection of every sentence, etc.

It's a staggering amount of work. I have the sneaking suspicion that Milo is going to be another one of those tech demo projects that quietly disappears after a while.
 
I still think it could work for a Star Trek game. It could understand phrases that you might say in the context of commands. Like "Fire torpedos" "warp 5" "shields up" "half impulse." You could maybe even point at the viewscreen and tell it to go that direction. In terms of ordering people around it might work. Maybe even for Rainbow 6 or such. The trouble is that you can tell Worf to fire the torpedos, but there's no way for them to script answers for any question you might ask him. Maybe he'll be programmed to talk about himself a little if asked, maybe he'll mention how he's feeling and maybe he'll notice if you're happy or sad, but the vast amount of work to charting it to the point you'd be convinced you were talking to a real person would mean that... it would take 30 years and it would just be talking to Worf and nothing else in the game, since it'd have to be entirely done with scripts. I think though that just scripting them to do limited, job related things would be great. I'm sure they'd even spend some time with Deanna Troi getting her to notice little details about you so she can "read" your mood.

I mean the tech is there to pull the information that the program needs. The interface is there. We're just not at the AI level yet.

Maybe over time scripts will build on each other. It's the input phase that's the biggest bottleneck. Take this for example for the simplest thing like saying hi to the machine, and getting it to respond... it needs to understand all the tons of different ways I can say hi. I can wave my hand, say "nice to see you," ext ext. So every year these things could get better and better at understanding us by just building on previously made inputs. The characters can then be differentiated by the different outputs they'd do as a response... while it might recognize 200 ways you say "hi" to it, it will only respond 2 ways, depending on if it considers your hi formal or informal... for example. So they won't have to repeat all this work for every NPC.

After 10 or so years of that, they'd probably get pretty good at understanding us, it'd be a question more of the painstaking effort for each npc to code all its scripts as a response. Also for references to a plot the character may be in... trouble. Big trouble. You're trying to interrogate this character for information. All they can really do is script around the line of questioning they imagine you doing. A conversation chain is filled with incredible, unending different combinations. The NPC has to know what you know, where you are in the conversation, ext. It could take forever and ever to do, or it would just look funny or they'd severely limit what you could ask it... which would be frusterating if you're trying to move on.

There's something interesting about the tech though that I am curious about. When I got the PS Eye for the ps3, I tried to use it as a headset... setting my speakers to output and the mic in the cam to the input. It works but it's unusable, because it picks up all the sound of the game and sends it to all my teammates, which they are not happy about. But it's cooler to just talk like you're having a conversation with your teammates on the TV than talking into a headset. If this thing is picking out your specific voice, does that mean it could function as a headset without transmitting anything but your voice?
 
Firestreak said:
I'm not saying there isn't a version out there that works. I'm just saying last night nobody knew shit... by now the video has been analyzed and people are sure it's scripted. This is based on there being a point in the fishing game where the screen reacts BEFORE her hands. Demonstrated with this picture, found by GAF.

9gx9p3.gif


Also by the fact that it knows her name, meaning the software was set up with that specific demo run through in mind.

This doesn't mean necessarily that the tech isn't as good as shown, it just means... well, it wasn't really shown to us for real so we have to trust them.

And it's now stated that it reacts to the tone of your voice, not really to what you say. Which means that the AI is NO WHERE near what was implied by the demo, but that has nothing at all to do with the Natal interface. I just am saying that it was partly the AI that really did me in, because Molyeux really laid in that talking up his creation to astronomical levels that I thought they finally cracked a new level of AI. It's just sort of a concept to show off the interface.

It's all well and good if you understand all this and examine the demo with this knowledge in hand. We can still captain the enterprise or play kings quest, only when we talk to people we're going to have limited dialogue options or else the're going to be saying, "I'm sorry I don't understand that." That's just an innate part of voice technology that based on today's technology and the technology for the foreseeable future that demo wasn't really admitting. Wen're not really at all further along than those old PC games where'd you type what you want to say to characters and spend 10 minutes trying to come up with the right line.
Click to expand...
I'm sorry, but what is that picture supposed to show?

The "HIT" circle shows a fish.
 
Uzor said:
I'm sorry, but what is that picture supposed to show?

The "HIT" circle shows a fish.
Click to expand...

No, it shows an impact on the water surface before her hand starts moving forward.

Of course, there's the possibility that the fish themselves can cause ripples, but I didn't notice any of the other ones doing it.

Moot point really, that kind of stuff is the "easy" side of the system. It's sophisticated software, no doubt, but it's obviously still a bit buggy/laggy/inaccurate from the demos we've seen so far. They can tighten it up with time though.

Facial emotional recognition is probably going to be the toughest part of the system's billed capabilities. I'd put basic voice recognition (given the inherent limitations they're obviously accepting) as an easier portion, while emotional recognition is going to be tricky with any sort of accuracy.
 
http://procrastineering.blogspot.com/2009/06/project-natal.html

New news

Btw, for those of you who may notice, this is the same dude that took the Wiimote and the Wii's technology to create that awesome 3D head tracking stuff :) Microsoft was smart in 2 ways:

1) Hired him
2) patented that shit :P which is why Nintendo hasn't made a game with that kinda technology, cause they'd get sued.

Also, my friend tells me Speilberg is very interested in making a game with this technology :P I'll have to ask him where he read that though.
 
You must log in or register to reply here.
Back
Top