none
Converting other joints to interactive zone coordinates? RRS feed

  • Question

  • I'm playing with the InteractionZone stuff, and I want to get the interactivezone coordinates of the wrist.  I'm sure this is a simple thing I'm overlooking somehow.  The sealed api returns the coordinates of the hand in this coordinate system, and of course I can get the depth coordinates of the hand, but the mapping between the two is unclear - for example, what what are the depth X coordinates of interactive zone value "0" and interactivezone value "1" values?  It's related to the kinectregion size as well, I guess.

    anyway - thx for any help

    Friday, July 19, 2013 7:23 PM

Answers

  • We don't have coordinate mapping from skeletal to Interaction Zone(PHiZ). The PHiZ target is not a static position so mapping after the fact is not an easy proposition.

    What exactly are you trying to do?  


    Carmine Sirignano - MSFT

    Friday, July 19, 2013 7:47 PM

All replies

  • We don't have coordinate mapping from skeletal to Interaction Zone(PHiZ). The PHiZ target is not a static position so mapping after the fact is not an easy proposition.

    What exactly are you trying to do?  


    Carmine Sirignano - MSFT

    Friday, July 19, 2013 7:47 PM
  • Hi Carmine

    I was interested in having the "grip" action cause a click, with the idea that it might be a more stable thing than the push-to-click (which is better then hover-to-click, but sorry it's still not there, imo).  Anyway, I got this to work by modifying the KinectButtonBase stuff, but there's still too much movement of the hand, so I wanted the tweak things so that the stock "hand" icon was following the wrist, and when the grip occurs, do the click wherever it happened to be (using existing machinery).  In just watching my wrist, it would seem like it is relatively stable when I do a grip... and it usually recognizes a "grip" by just bringing the handtip together with the thumb already... just think how nice it's going to be when the updated Kinect comes out where the hand tip and thumb are recognized explicitly.

    thx

    Monday, July 22, 2013 5:11 PM
  • Keep in mind that the Pressing action is the recommended action for this type of interaction. Each button interacts in a certain what with the action. Are you using the buttons from the Toolkit or taking into account the natural motion that occurs when "pushing"?

    Have you also looked the Human Interface Guidelines? A lot of research was done with respect to coming up with these and are the suggested recommends for Kinect for Windows.
    http://msdn.microsoft.com/en-us/library/jj663791.aspx


    Carmine Sirignano - MSFT

    Monday, July 22, 2013 11:02 PM
  • thx

    I was referring to the stock kinecttilebuttons and the default push-to-click behavior from the toolkit - I had assumed that some work had been done to deal with the "natural motion that occurs when 'pushing' " as part of the 1.7 update.  I have not looked into tuning that much under the hood, but it may be necessary based on my experience so far. 

    I am familiar with those guidelines and will look at them again. 

    btw, In a non-gaming space, I just don't think "we" have hit on the most appropriate scenarios and most effective gestures in those scenarios yet - Leap Controller's initial reviews are further proof of that, imo.

    Monday, July 22, 2013 11:16 PM
  • btw - I find the current pressing action in the kinect to also be suboptimal and flaky, and not an option for more than very very intermittent use.  It requires muscle tension across a large group that cannot be sustained ergonomically, imo.  I don't know if "press to click" will ever really work in a significant way, even though it is the first natural thing we'd think we'd want to have.

    The "grip recognition" feature, however, is slick and requires little energy, and will only get better in terms of requiring even less energy.

    Tuesday, July 30, 2013 5:42 PM
  • If you need other joints for interaction, why not just use skeleton coordinates?

    It's still possible to enable both skeleton and interaction streams, deal with skeleton coordinates, and using interaction for detecting grip action only

    Tuesday, August 6, 2013 8:28 AM