View my account

RealSense 400 Series Development firmware available



  • Jerry

    hi marty - with my case (detecting depth changes on a table top 2m away from the camera) which is the best setting ? 

    Do you think the changing the A-factor will have a beneficial effect on the RMS errors that are causing me ire with my hand-on-table detection ?

    that white paper link is broken btw - should be - missing a dash in there !



    Comment actions Permalink
  • MartyG

    Hi Jerry!   I will post links to your case with the table so that others can read it and know what we are talking about.   :) 

    Regarding recommendations for settings, some suggestions from an Intel support representative are provided here: 

    The above advice, and the sublinearity paper, recommends an A-Factor of 0.08 across the board.  According to the sublinearity paper, this setting increases average RMS error but smooths out small spikes.  Because it is the small spikes that have the worst effect on performance, gains may actually be achieved.  To quote the paper:


    This graph (below) shows spikes in the RMS values at values that correspond to Half-Disparities, which is why we also sometimes refer to this the phenomenon as the half-disparity issue.  As we use a more optimal A=0.08, we see that the overall average RMS value increases, but that the half-disparity spikes diminish or disappear.

    One may be tempted to think that since the RMS value is smaller overall for A=0 (ignoring the spikes), that this would be a preferred settings.  This is incorrect.  The smaller RMS values are a deleterious effect, also sometimes called pixel-locking.  Consider the extreme case of no subpixel behavior.  In this case all depth values would have to correspond to discrete disparities.

    When the actual distance to an object is between disparities, the disparity will lock to the nearest discrete value.  This value will clearly be wrong, but the RMS error could still in this case be incorrectly reported as near 0.  At the half disparity distance, depth values would suddenly become very noisy as they jump between adjacent disparities.  This is what leads to the spikes.

    In short, measuring a fairly uniform distribution of RMS Error as a function of depth is a sign of a good depth setting and camera.

    Comment actions Permalink

Please sign in to leave a comment.