bowling games is one that comes to mind that pseye cant do.... cant have a bowling game with 2d motion tracking. FACT.markinthedark
Technically true, but you also can't do a _good_ bowling game when you can't see if a player is twisting his wrist. Bowling without spin isn't bowling and doing weird alternative gestures to control spin isn't much better either. Actually, if you don't mind a bigger delay, you could do it in 2D. The arm might be 2D, but even from the projection ("growing" downwards then "shrinking", certain amount of sideways motion) you can get enough info to figure out speed and direction (would be much easier to trick, though).
For the time being the most accurate and "realistic" solution for bowling is a controller approach, since all the acceleration and rotation applied to the ball can be 1:1 measured from sensor data (and a camera helps knowing _where_ you are throwing that thing). I think this is true for most simulated activities where you usually interact with objects. It's almost funny how Kinect is going with a Harry Potter and a Star Wars game. Wait.. waving a _wand_ around? Fighting with a beam of light coming from a cylindrical object? Obvious choice for the one motion control device where you DON'T hold just that in your hand. Imagine sword fighting with a completely stiff wrist.
Also, seeing how even a commercial has people playing sideways for archery on Kinect is a big warning sign that depth might not be working too well. Or it's a workaround because your other arm might be hidden. Another factor with Kinect: if it can't be seen there is no fallback. I also remember people talking about boost not always triggering in Joy Ride, another hint that depth isn't working too well when your arms are in front of your body. If that part turns out to only work half of the time, it WOULD be just a really fancy and overcomplicated EyeToy 2.5D.
Well, thanks to the review embargo we won't find out until Nov. and after embarassingly failing live on TV it's possible we won't see anything but commercials on the upcoming TV appearances.
honestly the pseye can do very little of what kinect can do.
The difference isn't that huge. As hardware goes the RGB cam and mic array are actually slightly superior. 2D let's you detect edges and a players outline, 3D let's you determine his "front surface". The first is only workable as long as a player faces the camera and has his arms sideways or.. holds some kind of controller for which you can determine Z... oh wait. For the second it will be interesting how many issues with overlapping limbs they have fixed by the time it comes out. Some avatar glitches don't look like they even consider natural joint constraints (something I'd expect from any rag doll implementation these days). That would easily help with "is he crossing his arms of are they both broken and at really weird angles?".
Keep in mind that the camera as such only gives you a strange image with color coded depth. The entire "figuring out joints for skeletal animations" is purely software based. A limited version of that could be done with 2D images as well. Kung Fu Live almost goes there (it only puts bounding boxes around your limbs, ironically deriving joints from that wouldn't be a big deal). Too bad that "fighting" sideways is just about as awkward as shooting a bow sideways.
Well, eventually I wouldn't be surprised if MS adds a special "handheld sensor pack" (or at least a "split" controller for each hand) and Sony releases a "depth add-on" for the camera. Wonder what the PrimeSense camera would cost with its reference specs... or how annoyed the developers must be that their first "big outing" is with such a crippled version of their technology.
Log in to comment