First screenshots of actual Kinect 2 pointcloud data

#1

Here are some of the first 3D pointcloud results from with the new Kinect2 alpha development kit.

Higher rez versions (where you can actually read specs like resolution and field of view of the streams) on my site:
http://www.brekel.com/kinect2


#2

That point cloud looks pretty good, what are your thoughts?

I’m really stoked on using this new kinect to scan objects what you think about the possibilities there?

Thanks for sharing please keep us posted.

#3

Well to be honest I’ve probably spend more time in code than looking at the quality very closely :slight_smile:

But it definitely looks much more detailed and less noisy than the current generation Kinect/PrimeSense/Asus sensors.

It’s still a little early as it has only been out in the wild for about 2 days but it seems it brings improvements to all the areas it is used for.
Static object scanning, PointCloud recording, Body capture, Face capture, gesture recognition.

I’ll post more in the following weeks weeks while I progress.

#4

Very curious to see it.
I wonder how well this would work to scan larger objects like a car. Would make it easier on me to scan all our oldtimers.

But too bad the site is down.

will look later.

#5

Looks fairly detailed to me…

Makes one wonder if the sw could be abused to take a virtual peek inside people’s homes… :rolleyes:

#6

Unfortunately my webserver provider is dealling with some nasty DDOS attacks today so it may sometimes be a bit slow at times.

Btw any webcam or microphone can potentially be hacked and used to peek into someone’s home, the Kinect isn’t special in that regards. Just make sure you have security software installed that blocks any program trying to call home.

And if you’re really paranoid all you need is a piece of scotch tape and cardboard in front of the lenses. :slight_smile:

#7

Brekel, do you know if it distinguishes individual fingers?

#8

The fidelity is definitely there to see fingers but this is highly dependent on camera placement.

Doing a setup with the sensor mounted high up, looking down onto hands on a keyboard should be ideal.

Detecting fingers in a full body scenario would be more in the realms of detecting some hand poses. (open/close and maybe a few more)
Just because the fingers won’t be visible at times due to hand/body pose. Keep in mind that Pro mocap setups use high number of camera’s for exactly this reason.

#9

This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.