PrimeSense bought by Apple for $345M | Hacker News Reader

PrimeSense bought by Apple for $345M

geektime.com

137 pointsAviSchneider82 comments

This is likely a soft landing. Microsoft is actively developing their own time-of-flight depth camera in-house rather than using Primesense's "projected computed stereo" solution. Primesense probably saw the writing on the wall...

I like disassembling depth cameras and laser rangefinders. We used the sensors extensively in robotics, back when the only ToF option was the $10k Swiss Ranger SR4000. You can find some of my writings on my robotics website:

http://www.hizook.com/blog/2010/03/28/low-cost-depth-cameras...

http://www.hizook.com/blog/2010/06/20/low-cost-depth-camera-...

http://www.hizook.com/blog/2009/01/04/velodyne-hdl-64e-laser...

Can you recommend a low cost depth camera for robotics?

If you're indoors, any of the Primesense devices are best (Kinect or Asus Xtion). If you're outdoors, you can't really rely on the Kinect -- the projected IR light patterns get drowned out by the sun. Then you have to go a time-of-flight solution: the Swiss Ranger SR4000 (expensive!) or the PMD Technologies sensor (I've heard they're available for a couple $hundred now). Kinect2 is supposed to be time-of-flight when it is released too.

A small note to the HN moderators: the previously submitted title of this article mentioned Kinect by name, and I clicked through. If I had seen just the new title, "PrimeSense bought by Apple for $345M", I would have no idea what it was and moved on. I'm sure the new title is more succinct/accurate, but the old title was more helpful.

+1 especially given the fact the OP here is the Author as well.

This might be the last chance to get a developer unit [1]. Depends what Apple intends to do with the technology, I guess, but they usually seem to keep technologies for themselves. FingerWorks stopped selling immediately when Apple acquired them [2].

[1] http://www.primesense.com/developers/get-your-sensor/

[2] https://en.wikipedia.org/wiki/Fingerworks

There goes the industry. PrimeSense was, to my knowledge, the only company selling affordable room-range 3D sensors to independent developers.

There are countless unexplored avenues for 3D sensing technology, far beyond gimmicky gestural controls. My own application, as a high-accuracy presence detector for home automation, reached the front page of HN a few years ago (https://news.ycombinator.com/item?id=2303395). Others were using these sensors for DIY robotics applications, telemedicine, etc.

Now we'll probably have to wait another 20 years for another round of patents to expire before we "little guys" get another shot at depth sensing.

Good luck to PrimeSense, they probably deserve the exit, but bad news for the rest of us.

What about a kinect? They're capable, (relatively) cheap, and their ubiquity means that they have a good community.

Yes, but there won't be any new old Kinects, and the New Kinect requires entirely new reverse engineering and USB3, so it's not a drop-in replacement. My system uses Kinects, but we can't rely on them being available in the long run.

Don't know about he current device but Leap Motion can cover a football field. I don't think we'll need to wait too long and the tech will be even better and more developer friendly.

A football field? That would be quite useful. I got the impression from reviews of Leap Motion that it only worked in a small area directly in front of the sensor.

I think the current device was designed only for a small area but Leap's tech easily supports much bigger areas.

You're being too pessimistic. Depth sensing is not that complicated. Even vanilla stereo vision has gotten really good in the last few years and will only get better.

Can you explain to me how Depth Sensing works at a high level?

Some useful starting points:

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

https://en.wikipedia.org/wiki/Structured-light_3D_scanner

https://en.wikipedia.org/wiki/Range_imaging

Depth sensing as a small company in a minefield of patents is a lot easier if you buy your sensor from a company that has patents of its own.

What about the ASUS Xtion? We use them in our lab along with regular stereo cameras for DLT.

The Xtion is a PrimeSense design. We can hope for continued availability, but Apple's history suggests such hope may be in vain.

Unfortunate.

I've moved on to using DLT with 60-120 FPS cameras but I always root for depth sensors. The problem with the depth sensors is that the sample rate on them is just horrible (25-30 Hz which is unusable for any human motion that is beyond a stroll).

We'll see how it shakes out.

The New Kinect for the Xbox One is supposed to offer a much higher refresh rate, with less noise. I'm hoping that someone with the spare cash to buy a USB3 debugger will reverse engineer the protocol so we can use either the Xbox One Kinects or the next iteration of the Kinect for Windows on non-Windows systems.

Time to push open source software and hardware solutions. Design one, prototype and run a kickstarter.

nitrogen's posts indicate that he works in the field of mocap, and so he probably knows just as well as I do that these designs/developments are ridiculously hard to do.

Besides, kickstarters that rely on hardware fab probably have the highest failure rate on the whole site.

This purchase certainly makes one wonder what Apple is up to. Kinnect has been, for the most part, a party-game interface, and that's not really in Apple's bailiwick. Capri is not really small enough or appropriate for phones, since you tend to gesture with your phone rather than at it. It is small enough that it could be included in laptops and tablets though. There are obvious gimmicks Apple is likely to go for, like gesturing at your laptop to make it unlock or summoning Siri with a dance move. It might also be a decent couch-interface for Apple TV, if Apple hasn't completely given up on that yet.

Here's hoping Apple has something interesting under wraps to put Primesense to work on and isn't just filling up their patent war-chest.

I'd bet for the missing bigfoot, the Apple TV set (iTV or whatever they name it)

>Capri is not really small enough or appropriate for phones, since you tend to gesture with your phone rather than at it.

How about gesturing with your iWatch?

Just sayin'

> you tend to gesture with your phone rather than at it

Samsung at least have been trying to change that: http://www.samsung.com/us/support/howtoguide/N0000003/10141/...

Not that I've ever seen anyone do it.

Samsung never been succeeded anything else than copying something.

If you copy something, and the copy is better than the original in some way, is the copy still a copy?

When do you get to call it innovation?

When it's packaged and sold as an Apple product.

Did you buy the Galaxy Camera? That was the real first product that they can say something original. Unfortunately, nobody supported them. How about Galaxy Watch? It was the second, and also nobody helped.

Have you checked out Bada SDK? Have you used KIES manager?

Or, what's your point? Are you telling me that they really have something better? So they're not a copy?

In which way? Do you mean the Android which comes from Google? Because Android is just an assembly? Are you talking about [AMOLED](http://m.youtube.com/watch?v=z8I5oMxbO9Q&desktop_uri=%2Fwatc...). Or are you talking about TouchWiz widgets? Which looks so similar to iOS, even Google warned them to stop copying. Or Wacom tablet in Galaxy Note series? I know some people want it for Wacom tablet. Is it also Samsung's innovation just like Android which is also a Samsung's innovation?

It's irony that the word "Android" is actually a legally registered exclusive trademark owned by Samsung in Korea. Any other companies can't sell a phone named "Android" in Korea. So maybe you're right. That's very innovative strategy. But personally, I don't think real innovators won't do such crap.

Please tell me only one thing they really made better rather than copying other one. Not just larger and more. Because you won't call adding one more wheel to a Toyota truck as an innovation.

I am not talking they're not trying. They sometimes try, and any trial for original product horribly fails. Isn't it true?

They really suck at original or innovation. But for copying? Oh they're supernatural. They're doing it over decades, and even has been supported from government for that.

It's simply because you can't derive originality or innovation by copying others. If you ever have tried to make something new, or better, you should know that you must break existing one first. Copied but better? That's impossible doesn't make sense at all.

I didn't know that Google let Samsung register Android as a trademark in South Korea, that's pretty strange.

There's a story. Here're details.

An unknown Korean company registered the word "안드로이드" which is the Korean representation of "Android" as exclusive trademark in Korea, and Samsung negotiated with them and bought (or leased) fully exclusive right for the trademark from them. So other companies can't use the Korean word in their product in Korean market.

Google also registered trademark of the Korean representation for that but only in limited fields, (computing hardware/software), so there was no problem to register the trademark for any another fields (mobile phones, advertising, games communications, internet service, radio, TV …).

Maybe they're going to rework the Apple TV to target the market the wii was going after with iPhone/iPad game campatability on your tv? That is almost not a bad idea actually... Almost.

Based on things that I've heard from certain people, it looks like Apple is definitely going to join the console game market....

If this comes to fruition I just hope they get the controller right: it's of tantamount importance. Gesture based recognition for games is not something that makes up for a relaxing experience.

They tried once before, I wonder if they've learned anything.

Or is this like Apple is definitely going to make a smart tv?

"Here's hoping Apple has something interesting under wraps to put Primesense to work on and isn't just filling up their patent war-chest."

In almost all of the lawsuits I've seen, the patents used by Apple were filed by Apple (and most had Steve Jobs name on them).

Outside of the Nortel stuff, it doesn't seem like Apple is big about acquiring businesses for their patents. Are there notable examples of Apple weaponizing patents that way? I thought that was more the hallmark of Apple's competitors, rushing to put together a defense...

Looks small enough http://www.primesense.com/solutions/3d-sensor/ (see "embedded" photo at bottom). Apple can probably help miniaturize it further; also process shrinks.

New input devices open new possibilities. Just vaguing: 3D sculpting; with the device propped up, any surface is a keyboard; 3D photographs.

Ideally, couple it with a holographic display...

EDIT mobile applications "augmented reality gaming, virtual shopping, Real View™ measurements, 3D scanning and printing, photography enhancement" (http://www.primesense.com/market/mobile/), engadget article (http://www.engadget.com/2013/05/15/primesense-demonstrates-c...)

Remember the "Structure Sensor" on Kickstarter from just about a month ago[0]? That's basically a repackaged Kinect. Some commenter here on HN expressed his thought that 'this kind of sensor is something that will be definitely integrated with future smartphones, like normal camera'.

Kinect is not only about gestures. It's also a cheap 3D scanner.

[0] - http://www.kickstarter.com/projects/occipital/structure-sens...

Some things to note:

Laser diodes (which produce the structured light used by the PS technology) need to operate at a specific temperature so they hit a specific frequency of light. You do this so that you can hit a frequency window for the IR filter in front of your camera. The trade-off here is to use a wider filter, but this increases the system's susceptibility to noise and outside light (windows with sun shining through them are a big problem).

There is a longish delay for the laser diode to hit that frequency (that's what Kinect is doing when it's starting up; the firmware and all is ready to go in a few hundred milliseconds, but the LD needs tens of seconds to get to operating temp). It's not an instant-on device unless you can process data that has low fidelity.

In addition to dumping power into a heater, you also need a cooler (a mere heat sink won't be reactive enough). More power. I really worry about battery life in a hand-held device. (I don't remember how much power laser diodes draw, but the one in the Kinect draws over 60mw).

Microsoft has a metric buttload of patents in this area, both purchased and from in-house R&D.

I really wonder what Apple thinks it bought.

Apple TV + Kinect equals 60" iPad in your living room.

The application which makes most sense to me would be gesture controls for some kind of Apple TV device. Choosing menus in the living room, swiping between channels etc might do away with a conventional remote, perhaps.

I've been pondering on this lately. Apple certainly has set the bar with smartphones and tablets, making them practically indispensable. What next?

I don't have much expertise in OS's, but it seems to me that that IOS would be the perfect OS for smart machines. It's already optimized for mobile, wifi, cellular data, lightweight (relatively), and most of all, has a rich and vast developer community. Home Appliances that run on iOS would be a logical step, don't you think? In that case, this purchase would make sense. They could also be developing gaming units as expansion modules for AppleTV?

I'm really surprised Microsoft hasn't bought the company yet already. Kinect is the key differentiator for the Xbox versus the Playstation, and it can't be a positive thing for Microsoft's supply chain for a key manufacturer to be bought up by Apple.

Microsoft doesn't use PrimeSense for the XBox one. It's all in-house now. Should have negligible effect on the supply chain.

Presumably Microsoft has a perpetual license to PrimeSense's IP? Or does someone else entirely own that?

They bought another company in the field(3dv systems), which used a totally different physical principles(instead of dual cameras, using time of flight of photons) in their tech.

Canesta. I looked into their stuff before the Kinect came out. I am no expert in the field, but I thought their time-of-flight sensor is really clever. Basically, the speed of light is too fast to get an accurate ToF measurement within living-room distances using timers. So they use a technique where the intensity of the reflected IR is directly proportional to the ToF. Since you can measure intensity with much more accuracy and precision (and ease) than the time taken for light to travel a few meters, you get a pretty accurate 3D point cloud of the space being sensed. I can't find their white paper that explains this, but it's a good read.

That's interesting...was it an effect based on attenuation of the light? i.e. reflected photons from close objects will attenuate less than ones further away therefore intensity will be higher?

Something along those lines, but they also did something clever with the timing of emitting IR flashes and correlating the received number of photons with those flashes. I really wish I could recall more clearly, but I looked into it around 2007 - 2008. I tried finding the white paper, but all the links are dead :-( I might have a copy lying around, though... I'll try to dig that up.

Don't sweat it, I just think that if that's the method, it's a really cool hack.

Eureka! Found it on scribd:

http://www.scribd.com/doc/40735369/Can-Esta-101

Thanks for the link :) Do you happen to know how to download the pdf off scribd? It looks like you need to pay scribd to do it (or to read past the first few pages), which doesn't quite seem fair since I'm assuming they don't have a license to the content themselves, in which case it'd be someone else's copyrighted content they're profiting off...

Can't get it off scribd, but thanks to the magic of archive.org (seriously! amazing stuff), the entire pre-purchase canesta website is still up. Just look prior to mid 2010 or so. Here's the exact pdf in question:

https://web.archive.org/web/20091122234143/http://www.canest...

Yeah, that was a pretty big turn-off. Scribd is already on my groan-when-I-go-there sites, but I'm pretty sure a service that lets me download a pdf is not something I'm interested in paying for since that was invented decades ago when it was called ftp.

Doesn't return the intensity of reflection depend on the color and surface finish of the target? i.e. a white sofa reflects more light than a black sofa?

Yes, it does, so I'm not sure how (or if) they addressed that. I'll try to dig up that white paper.

IMHO, gesture navigation suck. It's much more expensive for your body to move your arm+hands around compared to one single click of a finger.

If you are playing a game, it's fine, you want to move around depending on the game style. But once you're looking for relaxing activities, like flipping over channels, is stressful.

> It's much more expensive for your body to move your arm+hands around compared to one single click of a finger.

It's less expensive than hunting around for the remote in the couch cushions for five minutes.

I agree and I think eventually we will stop fixating on the user input case. There are lots of other huge use cases like counting, tracking, modeling.

Totally agree. For playing games, it's a pretty cool experience. But for flipping channels, it's actually quite tiresome.

Regarding the headline, "A sensible exit: PrimeSense bought by Apple for $345M", does anyone have any opinion on the price here? It seems reasonable, but really just serves to highlight just how absurd the Snapchat offer was earlier this week (in my opinion).

I heard that PrimeSense wanted like $2B a couple years ago, from Microsoft. Sounds like they came down . . . :-)

If you still have a problem with snapchat you don't understand human behavior, startup dynamics, advertising and large audiences.

This is entirely true. I don't understand Snapchat's valuation in the slightest, which is why I mentioned it here in comparison to this acquisition. I'm yet to see how large audiences and advertising revenue are an answer to anything; see MySpace and Zynga, both of which fulfill(ed) those criteria. Snapchat seems like an easy service to clone in theory relative to, say, a YouTube or a Twitter-certainly Facebook could do it-so what are they bringing to the table exactly? Where does their value come from? A bit off topic here anyway. I just thought it was a nice comparison.

MySpace and Zynga are tiny compared to SnapChat.

This is disappointing (or encouraging, if you're an Apple competitor). Primesense is inferior technology. Kinect 1 (Primesense) uses structured light scanning, while Kinect 2 (not Primesense) uses far superior "time of flight" measurement.

And Leap Motion is way better than both.

Depends on the use case. Leap is not very versatile.

the leap motion is a good demo, but not practical to use.

Looks like Microsoft got the better deal here with a better camera and lower acquisition price.

http://venturebeat.com/2009/02/21/sources-confirm-microsoft-...

If Apple did indeed acquire Primesense, there certainly is good possibility that they will keep the hardware and software for themselves and stop licensing it out to other companies.

I wonder if this spells doom for the following products: ASUS Xtion http://event.asus.com/wavi/product/xtion.aspx

Structure Sensor http://www.kickstarter.com/projects/occipital/structure-sens...

This is probably bad news for leap motion who was surely hoping apple would acquire them. Maybe google will acquire leap?

Appple rarely pays full price and the non-apple opportunity is far larger. I'd say Leap remains in an even better position.

Didn't they already try to sell their tech to Apple before they went to MS, but got turned down?

I've worked for a number of years on the motion processing sensor side. The Apple acquisition makes sense, as it helps to add another input device for what I suspect will be an Apple TV. For the most part, everyone has hands and arms. To enable your hands to control what's showing on the screen seems in theory to be quite practical.

On the flip side, playing games is one thing.... controlling your TV is a different story altogether. I'm not sure how great an experience it will be. Samsung already has gesture controls on some of their camera-equipped TVS. The experience of waving your hands around for control can actually get a bit tiring.

It will be interesting to find out how Apple plans to integrate this technology into their devices.

Here are some facts:

    $85M investments so far.
    3 years ago, the company refuses $400M offer from Microsoft (rumour) - saying: 
        "we are here for the long run".
    It Powering XBox Kinect, Dumped at XBox One.

Don't ask me for citations, as this is a Israeli company, and the sources are mainly Hebrew tech-news sites.

Also amongst "Most Innovative Companies (2011)" [1]

[1] http://www2.technologyreview.com/tr50/2011/

I don't think this has anything to do with Apple TV or gestures on tablets. IMO, those are gimmicky applications of the technology that look cool but aren't practical on a daily basis. My guess is that Apple is making an augmented reality play, hedging that Google Glass will be successful and they'll have to compete with them. Infrared motion tracking has a lot greater potential to be successful in augmented reality versus the single camera setup Glass uses.

This makes a lot of sense especially if Apple wants to move Apple TV on. Lots of TV manufacturers like Samsung has gesture navigation implemented already.

This should be exciting.

For all intended purposes, this one is defintly meant for the AppleTV (or iTV). PrimeSense's tech is still years away from being able to put inside a smartphone (allthough tablets are another option which is more possible in the medium-term future).

This is a huge win for Apple! I expect to eventually see this in product categories they don't even offer yet, like smart home products. Not to mention the obvious, that they could add gesture support to the AppleTV.

I speculate this is a reaction to Google's recent acquisition of Flutter.

Reply on news.ycombinator.com