Kinect is not required to be plugged into the xbox one

[Scott_Arm]

I dont know how accurate the vgleaks number is from the above link but if so
60msec latency (16fps) is very noticeable lag, which rules it out for 'hardcore' twitch games.
This should be their number one priority getting this lag down as much as possible, perhaps reducing the cameras resolution etc will help. Until this happens kinect2 will just be a repeat of kinect1. Party/dancing etc games

16 fps? 60msec is basically 2 frames at 30Hz or 4 frames at 60Hz. From Kinect 1's 90msec, it is an improvement of 1 frame at 30Hz and 2 frames at 60Hz, which is appreciable, but the end result will still be some noticeable lag. Now, Kinect 2 skeletal tracking is being processed out of the System reserve, so there may be some wins be decoupling that processing from the game loop, in terms of overall latency in the game code from button press to result on screen.

Some twitch games already have terrible lag, as measured by Digital Foundry. Killzone 2, after patch, had 150ms of lag (not including display lag) between button press and response even after it was patched. GTA4 was measured from 133ms to 200ms. Those are two games people complained about. I'm not sure it's possible for Kinect to stay too far under those numbers. I don't know what the latency of the controllers was for PS3 and 360. 10ms or something?

It was not likely to ever be used for really twitchy gameplay. Dropping resolution will affect accuracy, which in my opinion was a bigger problem for Kinect 1. Of course, not all actions in twitch games have to be twitch actions. There is room for augmenting gamepad based games with motion control, facial recognition, voice control, if they keep the twitch controls on the gamepad. In a game like an MMO, even having your avatar mimic your facial expression, or being able to emote with your arms would be pretty cool. Those are pretty simple and straightforward things, but the nice kind of additions I'm looking for.

 

[Silent_Buddha]

Can you? If the tools aren't there to make it happen on a platform it might not be that simple. It might not even be permitted. Do we have that information?

That's a big potential problem, of course. Will Sony provide voice recognition libraries and dialect support like Microsoft? Or will developers have to roll their own. If it's the latter, then it would be highly unlikely for it to make it over to PS4 (like the voice commands in ME3).

Regards,
SB

 

[Silent_Buddha]

Some twitch games already have terrible lag, as measured by Digital Foundry. Killzone 2, after patch, had 150ms of lag (not including display lag) between button press and response even after it was patched. GTA4 was measured from 133ms to 200ms. Those are two games people complained about. I'm not sure it's possible for Kinect to stay too far under those numbers. I don't know what the latency of the controllers was for PS3 and 360. 10ms or something?

Much higher than 10 ms. Especially if you measure from button press to result on screen. IIRC while controller would average around 100-120 ms, Kinect v.1 would average around 133-150 ms in games. It wasn't that much worse than controllers. Reducing input lag by 30 ms would in theory bring it very close to console controller response.

The bigger potential problem for "hardcore" games will be whether it is possible for the camera to take input at the equivalent of 60 FPS like a controller, or if it is limited to taking input at the equivalent of 30 FPS. The latter wouldn't be a problem in most games for most users. But could possibly be noticeable to someone who is used to play COD all the time.

Although the argument could be made that it wouldn't matter anyway as no one (developer or consumer) would use Kinect as the only control method for "hardcore" games. As a casual input method or additional control method (controller + Kinect) the lower input rate wouldn't matter a single bit.

Regards,
SB

 

[XpiderMX]

I dont know how accurate the vgleaks number is from the above link but if so
60msec latency (16fps) is very noticeable lag, which rules it out for 'hardcore' twitch games.
This should be their number one priority getting this lag down as much as possible, perhaps reducing the cameras resolution etc will help. Until this happens kinect2 will just be a repeat of kinect1. Party/dancing etc games

I based my assumption on IGN hands on.

 

[warb]

As SB says there, it's not a controller replacement, there're a controller in the box. New Kinect can effectively track your upper half, or head and face to enhance control in traditional games. And there's still voice. A difference along with more accuracy and lower latency over Kinect v1, is that the resources for this are accounted for and thus at least available to every game.

Allowing XBone to function with sensor unplugged is sensible. Taking it out of the box would be a disaster. It's their USP. And we don't really want 2 identical consoles. I hope they don't internet focus group this system into the Xbox361 before we get to try it.


A claim for latency was about 2 frames less than Kinect 1 at the reveal.

 

[Scott_Arm]

Much higher than 10 ms. Especially if you measure from button press to result on screen. IIRC while controller would average around 100-120 ms, Kinect v.1 would average around 133-150 ms in games. It wasn't that much worse than controllers. Reducing input lag by 30 ms would in theory bring it very close to console controller response.

The bigger potential problem for "hardcore" games will be whether it is possible for the camera to take input at the equivalent of 60 FPS like a controller, or if it is limited to taking input at the equivalent of 30 FPS. The latter wouldn't be a problem in most games for most users. But could possibly be noticeable to someone who is used to play COD all the time.

Although the argument could be made that it wouldn't matter anyway as no one (developer or consumer) would use Kinect as the only control method for "hardcore" games. As a casual input method or additional control method (controller + Kinect) the lower input rate wouldn't matter a single bit.

Regards,
SB

Well, Killzone 2 and GTA4 and those other games had a full latency of over 133ms from button press to result on screen, but of that total latency of the gamepad itself was probably only 10-15ms, or something like that. Kinect 2 is going to be about 60ms. So the question I'd ask is whether they rest of the pipeline can be improved enough that the total latency for Kinect 2 can fall under 133ms, where 100-133ms was basically the best case scenario for 30Hz games last gen.

 

[Bigduo209]

Well, Killzone 2 and GTA4 and those other games had a full latency of over 133ms from button press to result on screen, but of that total latency of the gamepad itself was probably only 10-15ms, or something like that. Kinect 2 is going to be about 60ms. So the question I'd ask is whether they rest of the pipeline can be improved enough that the total latency for Kinect 2 can fall under 133ms, where 100-133ms was basically the best case scenario for 30Hz games last gen.

Well with Kinect it's a one-way street accounting for lag and reading movement. Controllers have the comfort of being a secondary hardware point for adjusting to lag with simple button responses.

But getting a response time (in motion-controls) similar to games using a traditional controller would have still have a disconnect. But maybe good enough isn't a bad thing. I think motion controls should be accurate and immediate, and it shouldn't be mandatory if the tech isn't as good as it should be. But if MS wants to help push it, then so be it.

I do wonder how they'll reiterate on Kinect after this version is released? Will they bundle Kinect 3.0 with the next Xbox, or sell that as a optional add-on during this upcoming console generation?

 

[Arwin]

DualShock 3 latency is 4ms. Move is between 4 and 21ms, depending on how many you track (4 max). Kinect 1 best case was brought down to 90ms iirc, and Kinect 2 now to 60ms I think. But it may well be that Kinect 2 skeletal tracking is much faster now, so that there is a big improvement to worst case.

Best button-to-screen update is using a ds3 with the XMB, which is 50ms all in. Best 60fps games do 66ms.

 

[Scott_Arm]

DualShock 3 latency is 4ms. Move is between 4 and 21ms, depending on how many you track (4 max). Kinect 1 best case was brought down to 90ms iirc, and Kinect 2 now to 60ms I think. But it may well be that Kinect 2 skeletal tracking is much faster now, so that there is a big improvement to worst case.

Best button-to-screen update is using a ds3 with the XMB, which is 50ms all in. Best 60fps games do 66ms.

The latency numbers for Kinect 1 (90ms) and Kinect 2 (60ms) are supposed to include the skeletal tracking, or at least that is my impression. The Kinect info on vgleaks says that is a measure of "end-to-end pipeline", or "60 ms with processing" whatever that includes. I imagine that is probably the best case.

They say latency has improved by 33ms, which would be 2 frames at 60Hz, matching up with what they said in the Kinect 2 talks and demos. I'd still expect a total latency of Kinect sensing to screen update of 133ms or more, if that was the average case for a 30Hz game on PS360 (100ms being the best for a 30Hz game). Kinect will more likely be a frame or two higher than that.

 

[zed]

As SB says there, it's not a controller replacement

Well that stance has changed a lot from the original promise from kinect1 remember that
'you are the controller"
I at the time said no, you will still need a controller to control movement shooting etc. you could use kinect2 for something thats infrequently used eg throw a handgrenade.

Im agreeing somewhat with zupallinere kinect is cool but not especially suited for games until its latency issues are sorted out

 

[warb]

Well that stance has changed a lot from the original promise from kinect1 remember that
'you are the controller"
I at the time said no, you will still need a controller to control movement shooting etc. you could use kinect2 for something thats infrequently used eg throw a handgrenade.

Im agreeing somewhat with zupallinere kinect is cool but not especially suited for games until its latency issues are sorted out

You'll occasionally be the controller for the games that track full body, but there's the option for "your spine as a third thumb stick" or head tracking for more traditional controller games. Voice should see the most use. Face and expression recognition is something that could find nifty applications.


I don't know what the latency is, other than it being improved.

 

[Shifty Geezer]

You'll occasionally be the controller for the games that track full body, but there's the option for "your spine as a third thumb stick"

We've had motion controls in sixaxis for a third thumbstick, and cameras for headtracking, for years, but barely anyone's done anything worthwhile with them. Ergo I place little faith in what could be. As I've said before, in MS don't show these fabulous features, people won't have a great incentive to mount a big block box on top of their TV. At the moment it's the family aspect, multi user instant login, and voice-control gimmick that justify attaching Kinect.

 

[LightHeaven]

I dont know how accurate the vgleaks number is from the above link but if so
60msec latency (16fps) is very noticeable lag, which rules it out for 'hardcore' twitch games.
This should be their number one priority getting this lag down as much as possible, perhaps reducing the cameras resolution etc will help. Until this happens kinect2 will just be a repeat of kinect1. Party/dancing etc games

Isn't 60ms less latency than any 360 or Ps3 30fps game? Or maybe even 60fps ones?

Even so, I think the biggest improvement regarding lag issues on Kinect 2 seems to be the far more precise data. Kinect 1 depth map was just too noisy, and it was the filtering/smoothing needed to have something acceptable to be used to animate the avatars that made the lag bad. Games or tools that showed the depth map directly, or even the kinect viewer tool (that applied the data directly to the avatar without any filtering) didn't have all that lag...

From what they showed of kinect 2 you only have to apply minimal to no filtering, which already reduces the lag.

 

[Cyan]

http://www.vg247.com/2013/08/16/konect-a-major-focus-in-unitys-xbox-one-support/

Kinect 2 is going to be a major focus in Unity's engine -the most popular engine among indie developers- for Xbox One.

Unity Technologies' CEO David Helgason defines it as an amazing piece of hardware.

Now that Microsoft is really into this whole indie thing, the folks behind the ever-popular Unity engine are working to make a version that specifically supports the Xbox One, much like they did with the 360.

And as they do this, Unity Technologies CEO David Helgason tells OXMUK, they’re going to make sure that devs are able to take full advantage of the system’s capabilies, with “a particular focus” on the new Kinect.

“Because it’s so powerful – it’s actually [laughs] I don’t know which kind of words to put on it – if I say it’s ridiculous it sounds negative, but it’s just an amazing piece of hardware,” Helgason said. “And the software behind it is actually a big part of that, the software that the engine has to support well to make this good. And so we’re working on that.”

Other Xbox One features he specifically mentioned that they’re working on are SmartGlass and the magical cloud.

 

[warb]

We've had motion controls in sixaxis for a third thumbstick, and cameras for headtracking, for years, but barely anyone's done anything worthwhile with them. Ergo I place little faith in what could be. As I've said before, in MS don't show these fabulous features, people won't have a great incentive to mount a big block box on top of their TV. At the moment it's the family aspect, multi user instant login, and voice-control gimmick that justify attaching Kinect.

Yes they should demo this stuff. Leaning slightly for Kinect could be way more natural than manipulating the motion of the controller. It could calibrate for varied sitting positions/slouching.

 

[Shifty Geezer]

The fundamental issue is developers not understanding natural input. Gamers, especially the non-core gamers, move in natural ways in response to gaming. Devs have had the opportunity to read these movements in various ways for much of this gen, but have failed to integrate them. An example I've used before is platforming. Many gamers will wave the controller a bit to try and jump higher in a platform, yet no platform reads that and actually makes the game do what the player is trying to do. And GT5's attempt at headtracking was dire, especially when you compare it to homebrew demos on YouTube. There is way, way more that could be achieved than ever has been, and as such I see little reason to believe any potential will be realised with any new technology no matter how inherently capable said tech is. Unless someone shows a real example of something working in full format, and not just a tech demo of what could be, I'm not going to value any potential feature of any system, and I expect a lot of internet-tracking gamers are similarly jaded. Even then, one or two awesome uses can still lead to no great changes in the long run. I loved Warhawk's motion controls but they were actually removed from the sequel, for who-knows-what reason - the existence of a game with good controls early in a platform certainly wasn't evidence of continuing adoption throughout the platform's life, even from the same developer making the same sort of game!

 

[Bigduo209]

The fundamental issue is developers not understanding natural input. Gamers, especially the non-core gamers, move in natural ways in response to gaming. Devs have had the opportunity to read these movements in various ways for much of this gen, but have failed to integrate them. An example I've used before is platforming. Many gamers will wave the controller a bit to try and jump higher in a platform, yet no platform reads that and actually makes the game do what the player is trying to do. And GT5's attempt at headtracking was dire, especially when you compare it to homebrew demos on YouTube. There is way, way more that could be achieved than ever has been, and as such I see little reason to believe any potential will be realised with any new technology no matter how inherently capable said tech is. Unless someone shows a real example of something working in full format, and not just a tech demo of what could be, I'm not going to value any potential feature of any system, and I expect a lot of internet-tracking gamers are similarly jaded. Even then, one or two awesome uses can still lead to no great changes in the long run. I loved Warhawk's motion controls but they were actually removed from the sequel, for who-knows-what reason - the existence of a game with good controls early in a platform certainly wasn't evidence of continuing adoption throughout the platform's life, even from the same developer making the same sort of game!

Same thing goes for racing games as well. How many times have you seen someone try to make a harder/sharper turn by moving their controller in a particular direction? Now how many developers have legitimately tried to incorporate that kind of functionality into their racing games or open world titles?

Only a few have done it on the PS3 where it's a viable option, and most (if not all) Wii racing titles have done it. But even then it's always done with varying success, like they didn't put too much into testing into it, or give an option to scale up/down the sensitivity levels.

So it's not quite the technology or it being a defacto standard that's at fault, but less of a willingness from developers to meet the tech halfway.

 

[-tkf-]

Kinect demo, you can clearly see the lag... i get an original kinect sense when i see that game...

https://www.youtube.com/watch?v=01y1vXBl1-8

EDIT: And before i am eating alive, the technology is really awesome, as can be seen after the Kinect Demo Game.. Now they just need to build a game that is as awesome...

 

[warb]

I didn't see a cursor. Are they using hand/eye for targeting? That'd be much better for pointing.

 

[Scott_Arm]

Kinect demo, you can clearly see the lag... i get an original kinect sense when i see that game...

https://www.youtube.com/watch?v=01y1vXBl1-8

EDIT: And before i am eating alive, the technology is really awesome, as can be seen after the Kinect Demo Game.. Now they just need to build a game that is as awesome...

The lag really doesn't look that bad. It's obviously improved over Kinect 1. It's still not super-twitchy, but it looks to be good enough for the types of control augmentation they were demoing. Tapping the side of your head to switch views is pretty cool. It frees up buttons for more important twitch actions. I'm not sure I like the targeting for the missiles, but I expect some other clever uses.