Leap Motion's not the household name Kinect is, but it should be — the company's motion-tracking system is more powerful, more accurate, smaller, cheaper, and just more impressive. Leap CTO David Holz came by the Verge's New York offices to give us a demo of the company's upcoming product (called The Leap), and suffice to say we're only begrudgingly returning to our mice and keyboards.
The Leap uses a number of camera sensors to map out a workspace of sorts — it's a 3D space in which you operate as you normally would, with almost none of the Kinect's angle and distance restrictions. Currently the Leap uses VGA camera sensors, and the workspace is about three cubic feet; Holz told us that bigger, better sensors are the only thing required to make that number more like thirty feet, or three hundred. Leap's device tracks all movement inside its force field, and is remarkably accurate, down to 0.01mm. It tracks your fingers individually, and knows the difference between your fingers and the pencil you're holding between two of them.
Holz showed off a number of different use cases for Leap Motion's technology. The simplest thing it can do is simulate a touch screen, so you can interact with any display as if it were touch-enabled — we were slicing pineapples in Fruit Ninja in seconds, without a moment of extra development or additional software.
Developers that do take advantage of the Leap's SDK will be able to do much more, however, and the possibilities appear to be limited only by your imagination. All kinds of different apps are being developed: some could improving remote surgery, others allow easier navigation through complex models and data, and others might put you square in the middle of a first-person shooter. It's like holding the Mario Kart steering wheel, but on a whole new level.
Rather than mapping particular gestures (cross your arms to close the app, draw a circle to open a new window), Holz said developers are being encouraged to provide constant dynamic feedback. No one needed to be taught what pinch-to-zoom meant — it's the natural thing to try and do on a touchscreen, and as soon as you start pinching or spreading it becomes clear what happens. That's the paradigm for the Leap, Holz says: you should always be able to just do something, and the app or device should respond.
Leap Motion's plans are huge (Holz mentioned a few times wanting to totally upend traditional computing methods) but the company's playing its cards close. The Leap will cost $70 when it's released — sometime between December and February — and Leap Motion is also working with OEMs to embed its technology into devices. The Leap is about the size of a USB drive, but Holz says it could easily be no larger than a dime, so adding it to a laptop or tablet shouldn't be difficult.
Developers are apparently beating down the company's doors for access to the technology — Holz said thousands of Leaps will be given away in the next few months, before it's released to the public. That's no surprise: after only a few minutes of cutting fruit, scrolling around maps and webpages, and navigating through huge 3D spaces, all without ever touching a thing, we're pretty sure we've seen the next big thing in computing.
The natural comparison to any motion control is Minority Report, an imagined future everyone seems to desperately want to come true. We asked Holz about the comparison, and if Leap Motion's technology meant we'd all have Tom Cruise's awesome PreCrime dashboard in the future.
"No," he told us. "It'll be even better."
Comments
this with windows 8, I can’t wait. Ill but the $70 day one.
By gambisss on 06.26.12 3:10pm
isn’t this article being premature in calling this device “more powerful” than the kinect? I mean all i’ve seen the leap do is at a short close range not like the Kinect that is a few feet away from the object it reads. Plus is this device tracking full body movement? I highly doubt it. So far all i’ve seen is hand gesture and that my friends is where the difference is in comparison with Kinect. Correct me if I am wrong but this is from what I seen and it’s quite premature to be advertising this device as a better product than Kinect.
By nikkiblaze on 06.26.12 7:09pm
I’m going to have to disagree with you. I own a Kinect myself, and although I don’t use it a lot I do know that the response times and number of vertices tracked aren’t even close to what I saw in this video.
Insane response times, near perfect object mapping in 3d space… and the way he was manipulating that molecule model? That does look even better than Minority Report.
By bkensey on 06.26.12 7:40pm
Apparently the limitations on the kinect are software based, MS said that they’ll be releasing firmware updates that will refine the kinect way past what it’s been doing up until now, I’d give you a link but I’m too lazy… google it….
By YozMan on 06.27.12 2:26am
Because Kinect was built with low-res cameras (a MS decision to reduce costs) no amount of firmware will change that. Kinect can’t read fingers and even if they could it would never reach the level of accuracy that the Leap is capable of.
I do agree that this isn’t tracking full body movements like the Kinect but you can’t convince me that Leap Motion doesn’t have a working prototype of this tech on an Xbox. Probably all wrapped up in legal red tape. That or maybe the Xbox firmware isn’t built to recognize such accurate gestures.
By mrlangston on 06.27.12 10:18am
Actually it can read fingers. There is an app in Kinect Fun Labs that lets you draw with your fingers. Also a ton of finger tracking things with the Windows SDK.
By DougB541 on 06.27.12 10:25am
I stand (actually I’m sitting) corrected. Shame on MS and developers for not using this feature more often but, playing devil’s advocate, does that speak to the unreliability of finger tracking?
It’s not as though using Kinect is an incredibly smooth and dependable experience. What usually takes up to 10 seconds using Kinect — waving to the device to re-enable it, finding the cursor, then long-holding your hand over a box — takes no time using the controller.
And yet I can’t bring myself to trading it in. Color me masochistic.
By mrlangston on 06.27.12 10:49am
I think part of it is limitations of the xbox360…its a 7 year old console…i honestly think that is part of the bottleneck. Probably only so much it can decide to track at one time.
For navigation i use kinect when playing a kinect game…don’t have to worry about finding and taking out the controller. But media i use voice….really works well and fast esp. once you know the line of commands…i can jump into netflix in 3 seconds.
By DougB541 on 06.27.12 11:35am
Kinect has been built with low-res cameras because of bandwidth limitations of USB2. I don’t know where you got that information from but they clearly don’t know what they are talking about.
If you want to “hate” properly, you could have said that MS didn’t include a processor in the Kinect to pre-process images (thus making it more efficient) to reduce costs. That’s a real fact.
By gagnon.eric on 06.27.12 7:28pm
This is not even comparable to the Kinect. It’s a motion sensor, yes. But it’s REALLY different for several reasons:
-The Leap tracks nearby objects only, so it’s immune to noise by design. Kinect has to deal with your dog, your furniture, people standing behind you, …
-Tracking a skeleton with all that noise isn’t an easy task. It requires a lot of CPU power to analyse the images provided by the Kinect. Tracking a hand isn’t exactly easy, but it’s easier. It doesn’t have to deal with “noise”.
-The Xbox hardware is kinda old and the resources are almost completely used by games. There is a complicated trade-off between accuracy and how much is left for games.
-The Kinect also streams an image feed to use by Games. Because it’s an USB2 device, this limits how much untreated depth information can be transmitted.
-I suspect they can use much more powerful infrared lasers since this isn’t projected directly into the users’ eyes. But this is only a theory. I know there are some regulations for infrared lasers but I’m unsure if you don’t have it if your device is pointed at the ceiling…
And anyway, Kinect 2 will have two sensors with small processors onboard, making it able to pre-process captured data and use stereo-imagery for near-perfect skeletal recognition. Basically it’ll be like having a motion capture studio in you living room, without the weird suits.
Oh btw Leap’s 3D object mapping isn’t perfect either. It doesn’t map the backside of the hand. That’s a problem with motion capture. There’ll always be a part that the sensors can’t see. That’s why the AI required for tracking is so complicated (slowing the response time).
I’m not saying that Leap isn’t awesome. It is! But to say that Kinect is inferior is like comparing apples and oranges.
By gagnon.eric on 06.27.12 6:40am
To show you just how the sensors are actually pretty good, take a look at this AMAZING video: http://techcrunch.com/2011/08/10/video-free-moving-kinect-used-to-map-room-and-objects-in-detailed-3d/
By gagnon.eric on 06.27.12 6:56am
I’m sorry, kinect is nowhere near as precise or responsive as presented demo. And the final product should be even better. Its in its own class.
By Formul on 06.26.12 8:56pm
yeah but a device that works within a foot of you will always be more accurate than one that works in about 7
By mrpakiman on 06.27.12 5:11am
Well, the Kinect senses a whole room and your body for the sake of basically kids gaming only. I’m pretty sure it could be like this too, if they wanted it to.
By baandoptager on 06.27.12 5:42am
If Kinect was just doing tracking of something, kind of, small then it would, probably, be just as accurate. Kinect is doing full body tracking. Not a fair comparison.
By Cinnamon267 on 07.01.12 2:53am
I’m pretty sure if you’ve done some development with the Kinect and used the SDK, the frame rates are pretty good but the tracking is laggy, 150ms in latency to be precise. While they claim they’ve reduced it to 100ms, it’s no where near as responsive as the Leap Motion.
By billyandika on 06.26.12 11:24pm
What is the latency of the Leap for fingers tracking? IMO 100ms is pretty fast considering how complex the algorithms for body tracking are.
By gagnon.eric on 06.27.12 7:31pm
I say… why not both?
By CobraCommandant on 06.27.12 9:20am
Sorry to say that but no one here seem to know ther real difference here. Kinect is camera and picture based recognition system, and it’s not accurate; and here it’s not based on camera, but i’m sure it’s base on ultra sound, like a sonar. And it’s is aproximately 1000 TIMES more accurate than kinect and it can used in the dark without light. It can really detect accurately the real shape of an object accurately without seing it. Kinect is not accurate and it need more power and more stuff to work. It is really an innovation here
By tsotsoa on 09.03.12 6:15am
This company needs to get big now
By patfactorx on 06.27.12 3:45am
Tie that with Google TV for the ultimate in remoteless remotes for controlling TV
By rfrost on 06.27.12 4:54am
who sits 3ft from their tv?
By scrapplejoe on 06.27.12 9:36am
Hiring professional robbers.
Task: Stealing the Leap Motion pototype along with the software and drivers.
By csdheeraj on 06.26.12 3:11pm
I know a guy who knows a guy…..
By tal4l on 06.26.12 11:50pm
I’m glad you guys are giving this the attention it deserves. I saw this a while back, but then it seemed like everyone forgot about it. It’s really crazy tech, and the price is IMO the best thing.
By Albanian on 06.26.12 3:11pm