New Kinect will sense more detail, heartbeat and minuscule shifts | Hacker News Reader

New Kinect will sense more detail, heartbeat and minuscule shifts

techradar.com

70 pointsfidz56 comments

If the rumors that the Xbox One won't power on without the Kinect attached are true, it'll be difficult to convince people to part with their used New Kinects for more interesting purposes.

I do look forward to seeing what can be done with the higher resolution, and especially, the higher field of view. My own application of the old Kinect, home automation, would benefit greatly from a wider FOV. Ditching the tilt motor was a good idea; I have to tell my customers to mount the Kinect without the tilt motor anyway due to stability issues.

I don't look forward to a day when everyone has Fahrenheit 451 full-wall screens (Illumiroom) combined with 1984-style monitoring capabilities (Kinect, PS Eye, smart TVs with built-in cameras). All the awesome things being done with modern technology and machine learning could be done on local CPUs without dialing home. I have to say I'm disappointed, although I'm not sure whether I'm more disappointed by technology companies for centralizing information gathering, or by the fact that my junior high librarian told me to read these books, making me aware of the dystopian possibilities.

Are you saying the new kinect offloads the image processing stuff to a cloud somewhere, like with Siri?

That's absolutely frightening.

This seems incorrect. The xbox has more than enough processing power to handle the tasks locally.

The new Kinect has been reported to have all of the processing it needs internally - and a 2 GB/s data link to push it all down.

I'm really excited to see what price point this comes in at.

No, that was just a general reference to current cloud-based trends.

> If the rumors that the Xbox One won't power on without the Kinect attached are true, it'll be difficult to convince people to part with their used New Kinects for more interesting purposes.

i read somewhere that they're, logically, planning to release a second-generation kinect for windows [1] soon after the console-version ships, which should alleviate that concern.

i always imagined that's what most people use for non-gaming purposes anyway, especially as the console-version uses a proprietary connector.

[1] http://www.microsoft.com/en-us/kinectforwindows/

The console Kinect uses standard usb and works with open source drivers. Kinect for Windows is also usb but requires Windows drivers.

thanks for the clarification; i wasn't aware of the driver distinction between the two. however, it seems we're both right on the connector part.. it originally used standard usb but has since been updated to use a different connector. from wikipedia [1]:

> Because the Kinect sensor's motorized tilt mechanism requires more power than the Xbox 360's USB ports can supply, the device makes use of a proprietary connector combining USB communication with additional power. Redesigned Xbox 360 S models include a special AUX port for accommodating the connector, while older models require a special power supply cable (included with the sensor) that splits the connection into separate USB and power connections; power is supplied from the mains by way of an AC adapter

..having purchased by console after the refresh i wan't aware it ever supported standard usb. it seems this will carry-over to the xbox one as well, given it too has a dedicated port for the kinect [2]..

[1] http://en.wikipedia.org/wiki/Kinect#Technology

[2] http://www.wired.com/gadgetlab/2013/05/xbox-one-photos/#slid...

People use the Xbox version of the Kinect because it's better supported by open source tools than K4W, and because it can be obtained used for very cheap. Adapters are available for less than $10.

The fact that we can get all this for a few hundred bucks sends chills down my spine. I think even the most diehare pessimists among us can agree that this is an amazing feat and that while we sometimes end up with "140 characters", we also get amazing things as a result of our efforts.

And something even crazier is that a hundred bucks today is nowhere as near valuable as one from even two console generations ago.

I wonder if it is using Eulerian Video Magnification[0] or something custom to do the pulse and motion detection.

[0] http://people.csail.mit.edu/mrub/vidmag/

you can detect heartbeats in infrared light easily, no fancy processing required -- see http://www.google.com/patents/WO2004078028A2?cl=en and http://www2.ece.ohio-state.edu/~anderson/Outreachfiles/Pulse... for examples.

Doesn't that mean it should be possible with the current Kinect as well?

For more detail on how the heartbeat detection works read "Eulerian Video Magnification for Revealing Subtle Changes in the World"

http://people.csail.mit.edu/mrub/vidmag/

Press 'A' to go forward, Press 'B' to go back.

[A]

You hesitated there. Are you sure?

... [A]

I can sense you're worried, would you like to talk about it?

Valve were looking at heart rate sensing as an additional input to their Director AI system (that adapts the entire game experience in Left 4 Dead and its sequel based on skill and other factors).

Look, I can see you're really upset about this. I honestly think you ought to sit down calmly, take a stress pill, and think things over.

It might help if you added the 2001 citation. Knowing the line was spoken by an actively hostile computer makes it even more relevant to the article.

Anybody else uncomfortable putting an always-on, always listening for a command, internet-connected device with a camera and 3D mapping device with this sensitivity in your living room? It's literally beyond the telescreen technology described in 1984. Cool - sure ... but jeez, it requires an awful lot of trust with whoever is on the receiving end of all of that data.

Yes I feel uncomfortable enough with my current Kinect that I have it disconnected until needed. However, thinking about it, Siri and Google Now isn't that far either. The leap towards this technology was eventually bound to happen and is unavoidable but what matters is where the data is gathered, processed and stored. Within the device or in the cloud?

I think the distinction between something like Siri and this is that this thing constantly has a live mic listening for a keyword, while Siri is completely reactive to intentional voice input. I guess Google Now is a bit more perceptually-aware, but that's based on data you've already implicitly granted them access to through signing up for their services and feeding those services data. I realize that aspects of this are the same - but video data of my living room and voice recordings of everything in my house feels FAR more personal to me than the emails in my inbox or something on my calendar.

It's not really a stretch to think that that capability provided by the XBOX ONE and Kinect 2 could be used for both good and evil. Same for Google Glass - it's always waiting for your "Ok Glass" trigger - but also processing everything else to see if it that keyword is said.

It's important to know where the data is being processed - but the courts have been pretty forgiving to companies that have overstepped their data-collection boundaries --- and even when they've had to pay a penalty, it's never come close to what I'd consider crippling to the point where they'd never make a similar "mistake" again if it were in the best interests of growing their business.

Yes. I'm not usually that fussed with privacy. Well, not as much as everyone else. But this is the thing that's stopping me wanting an Xbox One. Even if the logs were never looked at or opened, it's incredibly unnerving to know there is a 24-hour 3d scan complete with video and audio sat on a server somewhere.

> there is a 24-hour 3d scan complete with video and audio sat on a server somewhere.

i don't think that's happening. in its lowest power state, described as wake on voice, the device is listening (which is not the same as recording) for a prompt to power-on [1]. it validates what it picks-up locally on the device. only once it confirms (again, checking locally) a command to fully power-on, are other sensors enabled.

[1] http://www.polygon.com/2013/5/21/4353580/kinect-always-liste...

Thanks! That's a bit better.

It's not - but there's literally nothing technologically preventing it from doing that. It's one thing to put a device incapable of doing something into a room. It's something entirely different to put a device there with the promise that they won't abuse that power either by malice or error.

yes, but that's true of many other technologies as well. and while that doesn't alleviate such concerns, i think its clearly where the market's heading and (maybe naively) am not too concerned with it..

thinning back a few years, the idea that facebook could amass such volumes of personal data based on one's digital activity sparked much the same reaction. in deciding to use their service, there's a clear tradeoff between enjoying the benefits it provides and giving-up (and trusting them with) a lot of personal data - which, naturally, has potential for abuse.

technologies like this new kinect, as well as personal fitness monitors, google glass, smartwatches and other wearable technology, are just a continuation of that trend, now translated more to the physical realm.

it is definitely a leap, as the data being captured is more tangible and the ability to identify one based on it more real, but its still very much the same and feels like a natural evolution as we move more towards a world where everything is connected..

I share your concerns, but it's this also true of most laptops?

I keep a piece of tape over my camera, but there is nothing physically blocking the microphone.

One User Script you may want to install prevents the HTML5 voice input from being activated. It'll prevent HTML5-based browser voice-input if that's something you want, but it's worth considering if you don't need that function.

https://userscripts.org/scripts/show/110566

Don't forget heartbeat-sensing and detection of "miniscule [muscular] shifts", which combined with precise color analysis on your skin means the Kinect will be able to read a fair amount of the mental and emotional state we normally hide from the world.

I'm putting another tally-mark in my "Things Invented Towards a Cyberpunk Hell" column.

Why the double standard? Your phone has more sensors than the kinect and can determine your location trivially. Your laptop has a camera and microphone as well.

Its so weird to see this anti-kinect sentiment when we're surrounded by always-on internet devices with all sorts of sensors. The paranoid can just kick off the power strip or, heaven forbid, just not buy one.

I'm hoping this will won't be as accessible to the security services and hackers as normal communications seem to be. What are the chances our heartbeats and small movements won't be interesting to big brother?

When the red light goes on, stop having sex in the TV room. Or hey, install a Kinect in the bedroom and get a rating afterwards.

I'm pretty sure you could detect that with any old camera sensor, no pulse detection necessary.

Pulse detection doesn't need HD - it can work with fairly cheap sensors.

It sounds like it'll have higher resolution. Exciting to think what developers will do do once we get our hands on it.

I'm hoping there will be the same OSS code and drivers as there are for the current XBox 360 Kinect.

Huge boon for those of us with smaller rooms too - the Wired writeup making the rounds has a 6' tall guy standing 3' from the sensor and having room to spare.

That's my one gripe with the Kinect right now, it just needs too much darn space to work well as a gaming platform.

It definitely does have a higher resolution, an just as importantly it has lower noise because it has switched from structured IR to TOF. It also doesn't seem to suffer from the shadowing issues of the previous version.

Kotaku have a great video of it in action and walthrough the new features: http://www.kotaku.com.au/2013/05/kinect-2-full-video-walkthr...

You're saying it uses time-of-flight of light? Do you have more information about that because it seems awfully unlikely.

It does seem unlikely, doesn't it? Nevertheless it's true: we now have the ability to build cameras that measure the distance to each pixel by timing laser light bounces. Intel sells one too: http://forum.libcinder.org/topic/future-is-here-time-of-flig... http://www.interhacktive.org/1/post/2013/01/intel-creative-d...

The fact that that's possible at the scale of what appears to be centimeters is mindblowing.

Not visible light but infrared

Yeah, that's what I was thinking too- but apparently the tech's been around for awhile:

http://en.wikipedia.org/wiki/Time-of-flight_camera

the tech is around, but at very high price (a normal pointing laser costs 2k~3k, a proper lidar costs 30k to 70k). I think that is what the parent comment means "very unlikely".

The Intel Time of Flight SDK camera is $150 http://forum.libcinder.org/topic/future-is-here-time-of-flig...

Thanks! I am looking into this space for a while for my personal project. but 1m~3m distance is too small to my application. Do you know if this is plausible for direct sunlight and anyway to improve the distance?

PrimeSense (the company behind the Kinect IR technology) had a competitor at the time called 3DV systems (or something like that), that made an RGBD TOF camera that worked way, way better than the kinect - but they ran into financing problems, and were bought at-cost by Microsoft (something like $30M, which was the investment). That was in 2007, and even back then they were able to produce them for ~$300/BOM if I remember correctly.

At the time, everyone assumed MS just bought them to shut out competitors - but it appears they actually use the thing.

The company was Canesta. ToF sensors were their thing. [1]

Microsoft bought them in 2010, and it seems like they finally got their tech integrated (and cut Primesense out in the process).

[1] http://en.wikipedia.org/wiki/Canesta

And I was thinking about 3DV Systems' [1] product called ZCam [2]. I wasn't aware of the Canesta deal - apparently, MS cornered the market of depth cameras. Personally, I think this deserves some kind of regulatory anti-trust action. But that's just me.

[1] http://venturebeat.com/2009/02/21/sources-confirm-microsoft-... [2] http://en.wikipedia.org/wiki/ZCam

It's not explicitly measuring the time from emission to reception of individual pulses, that would be (pretty much) impossible with light. Instead, it sends out periodic pulses of light, and measures the phase offset of the lighting of resulting images.

How does using TOF instead of IR prevent shadowing? Unless it can see through you, it's still just going to have a single depth map producing gaps wherever it can't see, right?

A mistake I made in my excitement. TOF and structured light can both produce shadow regions of the depth map if set up like the previous kinect. What it looks like they have done is to switch the places of the RGB camera and the IR projector, which reduces the shadowing effect. Here's a Skitch of what I am talking about: http://i.imgur.com/y6VkRBm.png.

Slightly irrelevant, but I immediately thought of this CollegeHumor video sketch about a self-aware Kinect when I read the article: http://www.collegehumor.com/video/6452454/kinect-self-awaren...

you are always looking right before turning right...

The medical applications of this will be wide, with a heart rate monitor, the wide angle and being able to detect old people falling etc. I hope they release the PC drivers and sell a standalone Kinect 2.0 as soon as this hits the market.

I think standalone Kinect, compatible with Windows PC is coming. This will be huge, I agree.

http://www.theverge.com/2013/5/21/4353868/microsofts-new-kin...

Reply on news.ycombinator.com