HappySqurriel said:
In order to determine that a motion has occurred you have to take a trend of several points to determine what the motion was; if you require a trend of 10 points this would translate to a 1/3 of a second lag which would be highly noticeable and very annoying. Consider that the Wiimote captures motion at 100Hz and people found noticeable lag in motion capture. To make matters worse, no motion can be tracked unless it is long enough that it survives through most/all of the data-points of a caputre; and 30Hz would (potentially) mean that any motion shorter than about 1/4 to 1/3 of a second would be impossible to caputre. Obviously, this won't be a problem if you're tracking the absolute position of an object in space (as is occurring in the ball-kicking game), but if you were trying to play a game that's tracking motion it will be a huge problem. |
Where exactly are you getting that tracking 10 points equals .333 seconds of lag? Are you basing this off of 1 frame every .033 seconds that capturing at 30hz is capable of? The great thing about IR bursts, is that it captures the entire body structure at once, 30 times per second (if the 30hz is referring to the IR and not just the standard RGB camera to begin with). Since it is not tracking 10 points at their own separate intervals, the time combined to capture the motion will not be 10x the time for 1 point... With the dedicated processor doing all the wire frame updating and data processing, I highly doubt it would be anywhere near .333 or 333ms of lag. In fact, they have said many times now that the finished Natal product will be about 100ms of input lag which is actually better than quite a few shooters (Halo 3, Killzone 2 to name a few).
Essay Pro