I've seen the tech demo of that boy on the lake (which looks completely scripted). But I'm wondering what you can do with technology like that? It seems that you'll be running in a place most of the time and flailing about wildly. And just how accurate is the face, voice, and motion recognition anyway? Can this be successfully used in a game like Tekken, Call of Duty, Gran Turismo, FIFA or Ninja Gaiden? If not then what games would it be more suited for?

I think the answer is going to be: Wait and see.

I suspect things will be toned down significantly by the time it is actually released.

I think it is faked. Not that the example shown was staged, but that it is a lot less flexible than it seemed. The kid was unresponsive to a number of things the woman said. I think it has a very primitive kind of understanding, looking for keywords. While it probably works adequately most of the time, I imagine that when it fails it fails horribly. By that I mean it becomes extremely obvious how inflexible it is.

That said, I think even basic face, voice and motion recognition could be the inspiration for a number of cool games.

One thing I'm thinking myself: use traditional controllers but still monitor the players. Imagine what something like the Left 4 Dead "AI Director" could do if it was able to use the players physical responses to gauge tension. That kind of thing might make games far more interactive without relying on the players jumping around like lunatics.

Milo was such a smoke and mirrors bullshit demo it was disgusting. Lots of people getting really hyped up over a demo that's 90% fake and 10% misleading. It's not that there isn't cool concepts driving it, but they ignored those concepts in favour of trying to pawn him off as some revolutionary lifelike AI that you can have conversations with.

If someone had the technology to do that they would be billionaires, not doing tech demos at game conferences.

That said, I agree with rip-off that there are some cool uses that someone could theoretically get out of this. I also think that Milo could be expanded upon (read: actually made instead of faked) to the point of being a fun character to play with.

That said, I think we are years away from anyone actually intelligently integrating this niche add-on peripheral into any sort of game I would actually care to pay money for.

First impressions: Wow! That's amazing!
Second impressions: I wonder how that will work for real games; can you imaging having to interact/spend time with NPCs in a game? If you're into RP, it;s great, but for casual/non-RP gamers it'd be skipped...

The tech is amazing, but I also think that you'd have to have some serious AI to keep people's interest other than interactions with canned responses.

When I first saw Sony's remote I shurgged it off as a Wiimote clone, but as the demo went on - It seemed like it had a lot more of an obvious gamer impact than Natal does - I can see it being a lot more applicable to most games than Natal is... Natal is cool, but you need to wait to see it used for real to appreciate how it'd work - the Sony controller seemed more obvious

Natal != Milo demo.

That's simply one application of it as given by Lionhead, the fact they went on to show two more ways of using it pretty much shows that.

imo, the sooner everyone gets over the (of course it wasn't real) MS video and thinking that every instance of a Natal-enabled game == Milo-like the better we'll be.

Edit:
PS3 motion sensor thing... well, lets put it this way; we did a game on the PS2 last year where the camera tracked a foam green sword. The input might not have been 1:1 and not 3D (it was tried, but the resources required were pretty harsh on a PS2, most of our time was in the CV stuff) which the lights probably help with.. still, hardly ground breaking technology if that is all it is. About the biggest win they have is that the PS3's Eye is better than the current 360s and the PS2s which, combined with known light, probably makes it easier.

Of course Milo isn't intelligent- that's not the point. The point was to show that the new system apparently:
-Has facial identification and recognition
-Has voice identification and recognition
-Can isolate the human body from the rest of the playing room
-Can accurately track the body movements accurately and in real time.
-Can be programmed to identify other objects from the room (eg, skateboard).


Did Milo really see the picture of a fish or did he just see a white thing blocking the view? What is the resolution and response time of the camera?

I would find it hard to believe that any kind of advanced recognition system is even possible in real time with current tech, but I was impressed by the possibilities.

Quote: Original post by GMuser
Did Milo really see the picture of a fish or did he just see a white thing blocking the view? What is the resolution and response time of the camera?

Apparently he saw the paper, scanned it and recognised the fish...

I think that the PS3 controller will have the most immediate impact - to me it's like the Wiimote on steroids, should finally allow sword fight games, etc...

But the Natal stuff... it's something you put your long term hopes on, it;s got the face recognition stuff and the full 3d spacial awareness (eg: arms, legs, etc), but I'm pondering how it can truly become the common interface to games... maybe if they used that object recognition tech to say "ok this is my gun" it might be cool, but I think for quite a few games you need the tactile trigger nature of the PS3 motion stick thingy.

I really want to see the potential for both of these and actually hope to see some convergence in the future.

Quote: Original post by GMuser
Did Milo really see the picture of a fish or did he just see a white thing blocking the view? What is the resolution and response time of the camera?

Keep in mind that Milo is a project which has been on the go since 2001, previously known as The Dimitri Project, so they have had plenty of time to work on image recognition and other things in that time. As to what Milo saw, well maybe he does just see a white thing, it's hard to say without someone drawing something which looks nothing like a house.

Quote:
I would find it hard to believe that any kind of advanced recognition system is even possible in real time with current tech, but I was impressed by the possibilities.

I'm not sure how much of what I know is covered by NDA, but given what I've seen done on a console and the fact this has custom hardware based on another product I've seen I wouldn't be at all surprised if it is as good as was demo'd live on the stage if not better.

I have two problems with natal and similar tech as is:
1) Walking. You can't really walk properly. You would really need like a treadmill or something that you could actually walk on, without actually moving.
2) You have to look at the screen. This could work if you had those 3d glasses.

I would imagine that most games will still require the controller. The walking thing isn't major (and probably wont happen in this gen of consoles), but the glasses I think would really add to the experience.