Hacker News Reader

Comment by mrsebastian | Hacker News Reader

Any idea what the tech behind it is? They're very vague on the website, and my patent searches are turning up empty.

I guess some kind of infrared thing?

I suspect infrared time-of-flight. http://en.wikipedia.org/wiki/Time-of-flight_camera

You can get pretty impressive accuracy and precision at short distances, and the plane of depth points matches what their demo video shows: http://www.youtube.com/watch?v=_d6KuiuteIA&feature=playe...

I've read some other comments claiming that they can cover football stadium sized venues. If that's true, I think that suggests laser illumination.

Yeah, my guess is some kind of LIDAR type device and perhaps a mix of a couple of other things. A researcher friend of mine was musing over possible uses of LIDAR a couple of years ago and this was one of his end uses if my memory is correct.

I could of course be wrong, so if someone knows different

I thought LIDAR was prohibitively expensive technology for use in consumer applications, but I could be wrong. I remember reading about it when Radiohead did their "House of Cards" music video with LIDAR tech.

Some relative of Kinect if I had to guess.

Kinect is using stereo-vision using 2 cameras...I don't think they have the same technology here

Yeah, no. Kinect uses an IR emitter with an IR-based depth camera, there's nothing stereo about it. A regular RGB camera adds color, but that is all.

Not to nitpick, but there is a stereo aspect to it. The IR projector is offset laterally from the IR camera, and this is critical to evaluating the depth, because the disparity of the IR dots is more extreme the greater the distance between the emitter and the camera.

It is actually exactly stereo, except one of the cameras is replaced with an IR projector; you basically do the standard stereo math as though the projector is a camera 'seeing' the image that is being projected.

you must be thinking of something else. kinect has two cameras, one for laser IR particle-filter type stuff, and then an rgb camera. it isn't stereo vision with two regular 2d cameras, however, doing displacement math. it is looking at the grid of dots projected and seeing where the dots moved in order to create a depth map.

Reply on news.ycombinator.com