[Lab] MS Kinect question

j ross waterfallclose at gmail.com
Wed Aug 14 13:30:49 EDT 2013

Thanks Jason, I appreciate the feedback - sounds like I should order the
Kinect for Windows if just to avoid connector issues (+$200 is an expensive
connector though - 2x the price of the sensor seems a bit much, even for
MS?  :).

I'm just assembling a new PC so hopefully everything is up-to-date (Win8 -
unfortunately) I opted for a new build rather than a laptop so I could have
the option of multi-sensor in the future.
Where did you order your Kinect from?  I'm currently looking at Techdepot
but would appreciate any buyer advice - I haven't ordered from them before.

BTW - room scanning is a nice-to-have, not a strict requirement at this
point.  I'm more interested in people scanning and gesture tracking (for
automation control) at the moment.
Any thoughts on Artec Studio?  The ability to auto-create a "watertight"
mesh is attractive, but pricey.

On Aug 14, 2013 12:56 PM, "Jason Cobill" <jason.cobill at gmail.com> wrote:

>    I'm just in the middle of a bunch of Kinect experiments myself, so good
> timing! :)
>    The Kinect for XBox has a non-standard connector to plug into the XBox
> - with the Windows version you get an adaptor that converts from XBox to
> USB, and plugs into the wall to power the Kinect camera.
>    I've seen some really cool demos on the Scanect software site regarding
> environment and people scanning - James on this list has some more direct
> experience with that software. I've been using the OpenKinect API on the
> Mac to access the data directly - there's some pretty rad stuff going on in
> there. :) You can also yank pointcloud data directly into Meshlab now using
> the Kinect/OpenNI plugin for Meshlab, but it only works on Windows 7 & 8 -
> hope your OS is up to date! I'm running Vista and had endless headaches
> trying to get the hardware to talk to the software.
>    A side note about scanning environments: The depth data from the Kinect
> is logarithmically scaled instead of linearly, so the further you get from
> the camera the bigger the error gets. Up close (torso scanning) you can get
> around 25points/cm^2, which is fantastic, but I think if you're intending
> to scan rooms you might want to go with a more sophisticated lidar tool
> that doesn't suffer from so much distortion. (Although, as I said, the
> Scanect people have an impressive demo of a small room being scanned, so
> it's not impossible)
>    -Jason Cobill
> On Wed, Aug 14, 2013 at 12:42 PM, j ross <waterfallclose at gmail.com> wrote:
>> Hi all - I'm considering getting a MS kinect, but wanted to get some
>> opinions and advice first:
>> - difference between MS kinect for windows and xbox (eg.
>> performance/features, OpenNI compatibility)?  (They appear identical but
>> xbox ~$110, windows ~$270 - SDK seems free so I don't understand the
>> difference except maybe commercial use markup)
>> - where to buy MS kinect for windows in Canada? (I only see amazon as a
>> supplier 3-5weeks lead)
>> - if xbox (eg. from best buy), what else do I need to connect to a PC?
>> Do I need to hack the connectors? Does the connector adapter come with it?
>> - comments on Xtion Live, Carmine sensors?
>> I have googled these topics but there don't seem to be clear answers.  My
>> main interest is for scanning people, large objects and rooms as well as
>> future work for gesture input.  I'm currently working with David Scanner
>> for small to medium static objects with structured light - takes too long
>> for people/rooms though and not suitable for gestures.
>> Thanks!
>> Jeff
>> _______________________________________________
>> Lab mailing list
>> 1. subscribe http://artengine.ca/mailman/listinfo/lab
>> 2. then email Lab at artengine.ca
> _______________________________________________
> Lab mailing list
> 1. subscribe http://artengine.ca/mailman/listinfo/lab
> 2. then email Lab at artengine.ca
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://artengine.ca/pipermail/lab/attachments/20130814/c72caa78/attachment.html>

More information about the Lab mailing list