Also, it's interesting that someone mentioned Sixaxis. In theory, Kinect v2 can replicate pretty much everything that Sixaxis can do combined with a standard Xbox One controller, albeit likely with a bit more lag.
Regards,
SB
I doubt that. You're going to have a whole lot more issues with that. There are many subtle movements you can do to the Sixaxis that will register properly but probably couldn't with the kinect due to the resolution the Kinect can offer.
With some research Kinect V1 appears to have
Depth resolution: ~ 1.5 mm at 50 cm. About 5 cm at 5 m.
Which is expected, the farther away, the less accurate.
I think the main issue with kinect and implementing it in stuff is two things:
First is the 60ms delay, which is ~2 frames in 30fps games, and a 3.5 frames in 60 fps games. This delay occurs even before you factor anything else, so that is a concern. (which many have chimed in about)
The second is how you implement it. If you simply just throw the skeleton data into the game it would work very nice. (resulting in games like rock climbing which works pretty nice in my opinion)
However, the majority of the games simply cannot support the player 1-1ing the character in game, and as a result you need to use "gestures" to inform the game that the player has thrown in some type of input. This is hugely problematic in that you're trying to strike a balance between two issues that can make games borderline unplayable:
1. Input recognition lag. What is your criteria for a "gesture"? In all cases, you REQUIRE the player to complete at least a portion of the gesture to be able to register the gesture itself.
However, you already have 60 ms Kinect v2 lag + rendering time + any other shit like display lag, and tacking on a further, oh I don't know, lets take 3 frames from the 30 fps 1080p camera from kinect v2 to properly recognize a gesture, you're looking at a 100ms (3*33ms) + 60 ms + rendering time + any other shit like display lag, This number may easily blow past 200ms.
2. Which is connected to the idea above, which is false positives. If you want to properly recognize movements without delivering false positives, (providing that you use the same code of course) you have to extend the number of frames you use to detect gestures, more frames = more lag.
And yes we are seeing these issues still present in Kinect V2 in the currently shown games. The X-ray FPS game where you can hold up a shield is a clear example.
Gesture recognition is clearly not the way to go in games if the game itself is intolerable to input lag. You're much better off ignoring it entirely and sticking to buttons to avoid a minimum of 100s of ms of delay.