Kinect technology thread

Not only that, most people have coffee table between the TV and couch. e.g., Kinect may have problems seeing the foot movement for acceleration/brake in these cases.

It's a much more complex tracking exercise in practice. Demoing on the show floor may not be indicative of the real user environment (e.g., couch, cushions and sitting/leaning bodies are "fused" together, plus stuff on the table).

 

That’s why I think that almost every game that will support sitting down, will depend on hand movement and voice command, in the case of Forza Kinect demo the game accelerates and brakes automatically.

 

Yes, and this is where it may get sticky because of the 320x240 depth resolution. The couch is usually further away than stand-up-and-play distance. So the hands may appear too small to the camera. Need to involve very clear hand movement (e.g., open palm vs fist, large swing).

If they implement other form of recognition using the video camera (e.g., 640x480 @ 30fps), then it may be sensitive to lighting condition and gesture speed.

They will need to tune the UI carefully. Little kids have small hands, which may or may not be a problem depending on the target audience.

Select voice commands will work in certain context.

Ah, I was probably thinking of the driving game they demoed (using foot pedal to brake).

 

Yeah, that was the early Project Natal Burnout prototype demo they showed last year. None of the announced racing games(Joy Ride and Forza 3) use your feet that I can tell.

Tommy McClain

 

Is there video of this? I'm wandering if people were actually sitting on a chair or a stool that has nothing in the back.

 

I haven't seen any. All videos were either at the Media Briefing or in the Kinect demo pods at the Microsoft booth. Neither of which had anything to sit on. I believe the comments about sitting while playing Forza 3 were just comments from Turn10 that it can be done.

Tommy McClain

 

I would be surprised if producers work by simple assumptions. Like the aborted finger game, they would at least confirm it via quick discussions, available documents or/and prototype.

It's probably too high risk, and too late to meet his schedule and budget.

EDIT:
There you go... stolen from GAF:
http://www.neogaf.com/forum/showpost.php?p=21911396&postcount=1

Kotaku's update:
http://kotaku.com/5565777/xbox-kinect-does-not-play-well-with-couch-potatoes

 

That might be true, but does that mean the feature isn't being implemented in RUSE for XB360 because of a misconception from the producer, where it'd actually be a viable and worthwhile addition that could help shift units? Or are they actually working on Kinect integration and the guy just got his story wrong?!

 

Obviously, MS is in full damage control on this point. There's nothing quite so amusing as making a self serving statement like, "games will work for seated players with Kinect if that's what the developers design for" when they are simultaneously telling all their developers "only design games where the player stands". The subtext is, of course, "we can't get Kinect to work well for people sitting down." I imagine the Forza sitting demo was probably accomplished with a tall stool in an optimal position so Kinect just thought it was looking at a standing person. That's a pretty different version of "sitting" than most people will have in their home where Kinect is going to be seeing your whole couch, armrests, an ottoman, cushions, etc, etc.

 

There's a danger of reading too much into an answer.

As the RUSE developer, his only point-of-view is "would Kinect would work in a useful manner for RUSE?". I'm guessing he asked his development team that question and they said something like "it wouldn't work whilst seated", so he said "ok, don't bother".

So when asked a question, he answered truthfully that from his point-of-view it wouldn't work whilst seated.

Such an answer could mean a number of things:
- the Kinect/seated profile will not be adequately developed by the release date.
- the Kinect/seated profile would not support key functions of RUSE.
- making Kinect/seated work with RUSE would be too much work.
- Kinect/seated is not "licenced/recommended" for certain game types.
- it was Friday afternoon, the developers were off to the pub and the cleaning lady told the publisher what her friend read on the internet.

 

Ah, an interview:

http://www.nowgamer.com/news/3658/ruse-dev-explains-kinect-snub

Annoyingly, the producer isn't quoted as saying the interesting bit - that seems to be either a conclusion of the journalist, or something said 'off-camera'.

 

The first point says MS isn't ready for seated gaming, which is what people are saying, and where they hope to have seated gaming working in all instances, there remains a question mark to what standard they'll have it working by launch and later into the product's lifecycle. Importantly if it's in a poor condition at launch, there's be little support from the core gamers and the platform risks being relegated to EyeToy like party games.

The second point is invalid as Move works while seated, and all they'd need with Kinect is to replace Move pointer movement to arm pointer movement.

The third point says MS's libraries aren't in a very useful state, which is a variation on point one and how much integration in traditional games Kinect will get

Point four only exists if point one or three is true. There's no way MS would prevent use of Kinect input when seated just to force all games to have to be played standing up!

So far, what we've heard points to seated Kinect tracking not be very functional are the moment, and I haven't heard or seen anything to the contrary. The nature of Forza's seated presentation is the one main contrast to this view of the current viability for Kinect in conventional, seated, core gaming. I'm sure if they get it up and running in time, RUSE will have it included, as they can take the same interface from Move and just replace the input code with a little tweaking.

 

Well it's much the same thing. What's stopping Kinect gamers sitting in a chair to play an RTS if not technological limitations? Either the technology doesn't support sit-down gaming, or MS mandate no sit-down gaming, artificially limiting appeal of their platform, or the developer won't support Kinect sit-down gaming because they feel no-one who buys kinect is going to want to play this game as everyone buynig Kinect is only wanting active gaming, thus limiting the appeal of their title. The latter two options seem like poor business to me, hence the former, along with other rumours and reports, seems the most likely.

Edit : the guy says, ""Kinect is more for interfaces where you are standing and making more immersive gestures..." The question is, why is Kinect for those sorts of interfaces?

 

Most of what Kinect does is in software, not the hardware. If sitting is a limitation of the actual hardware because of resolution or the IR patterns, then I'd say it won't be fixed. If sitting is currently a limitation of the software, because of the way they interpret the data, then it could possibly be fixed. I can see it being either one, if not both. Hopefully it's a software problem.

 

That quote could mean several different things. When he says "is more for", does he mean technologically or does he mean practically?

 

Yes, but I already explored that a couple of posts up. who is limiting Kinect to stand-up games? This developer who doesn't see a market outside of stand-up games; MS who doesn't want anything other than stand-up games; or the technology not (yet) able to integrate sit-down gaming?

The former two options don't make business sense to me, the developer in this case reducing appeal of their title, and MS reducing appeal of the platform in the second case. There's an argument to be said that they are trying to differentiate the platform and for marketing reasons are distancing Kinect from core gamers, but I don't see why they would force developers to avoid sit-down gaming; they'd only need focus their marketing campaign on the stand-up experience to communicate a clear message to casual gamers.

 

Kinect is not targeted at core gamers - the device isn't really suited to that type of game.

Not exactly:
- in this situation Move is "dumb pointer", not a motion controller. There are no facts (afaik) about the exact accuracy of Move, but I believe the motion+ is "accurate" to movement of 0.1 degrees or less? and it's rumoured that Move is more accurate than that? Assuming 0.1 degree, that would give 600 easily accessible "pixels" over 60 degrees (30 degrees per side).
- kinect has a total vertical resolution (ceiling to floor) of 200 pixels, due to the light technology there is no way to 'zoom'.
- Move, being a bluetooth dumb-sensor sends the rotational data directly to the PS3 along with the button states, it should be relatively trivial to read that data.
- Kinect creates a depth map, that is sent to the 360 and processed by the libraries, according to the "Kinect sports" developer this takes around 10-15% of the "total system resources".

On the face of it, implementing Move as a "rotational mouse" is stupidly trivial. Doing this with Kinect is far from trivial.

I doubt RUSE would implement it, but I guess other games might if it can work reliably.

 

The move devs mention "subpixel accuracy" in their videos. I don't know what that actually means though.

 

I seem to remember the Move being about 1024x1024 pixels precise for pointing on the screen initially, but I don't know where the precision is for the final driver and model. Precision for turning the device over its various axis is around or just under 1 degree I believe, and I vaguely remember there being a figure quoted somewhere for z precision as well.

 

...but Move has what we in the future like to call "buttons"