Camera Priors for turntable scan?

Comments

4 comments

  • Avatar
    Götz Echtenacher

    Hi Justin,

    why do you think RC ignores certain settings for different aquisition techniques? There is no way RC can figure this out by itself - it's not intelligent enough (yet) !  :-)

    I would think that epty gps values are ignored, at least I never had any problems with this (that I am aware of).

    I don't know too much about priors, but I am pretty certain that they are used as starting-off point, so it might very well be a bit faster but also less prone to come up with an entirely wrong undistortion.

    0
    Comment actions Permalink
  • Avatar
    Justin Goode

    haha, not saying that RC ignores certain settings based on acquisition techniques.  I just want to disable any settings were created and intended for use in aiding alignment of laser, or captures where you're physically moving the camera each shot , in order to reduce any extra computation that would be useless for a non moving camera.  Camera priors, for example, might be one of them since the camera never moves.  

    From what I understand, using a turntable is sort of tricking photogrammetry programs into thinking that the camera has been moving around.  By default, it seems all of these programs are built with the assumption that the camera is physically moving position, thus many of the settings were created with that in mind.  

    I've been trying to go to the help menu for each setting and while it technically tells you what each setting does, it doesn't always give you an idea of when/how it's beneficial to use them--"yeah, if your camera is stationary, this setting is pointless as it's intended to help align moving cameras and laser data--If your camera is stationary or not laser, leaving it on just causes RC to go through extra calculations that it doesn't need to and is slowing down your calculation" <<< that is what i'm better trying to understand.  

    0
    Comment actions Permalink
  • Avatar
    Götz Echtenacher

    Hehe. Sorry, I misread your question! Duh.

    But I understand that you expect certain values to only make sense for certain aquisition methods. I think GPS is the only one that really matters and can only lead to problems IF it is present in images AND contradicts the coordinate system from other means of geo-referencing (e.g. laserscans in local systems).

    This is really a relativistic question. You seem to expect there is a difference whether the camera moves through space (relative to the earth) as opposed to the camera remaining still (relative to the earth). But that really doesn't matter. RC only cares for the movement of the camera relatice to the OBJECT. So if you use a turntable or move through a building with the camera has no impact on the way RC is calculating. With turntables it only (regularly) leads to problems if the background provides too many tie points and RC "decides" that this is the intended surface. Hence it is the good practice (and challenge) to chose the surroundings so that there are as few as possible (or even no) tie points at all stationary (relative to the camera). Does that make sense at all?

    0
    Comment actions Permalink
  • Avatar
    Justin Goode

    Gotcha.  Yep, that exactly answers what I was trying to understand.  Thanks!

    0
    Comment actions Permalink

Please sign in to leave a comment.