Missing cameras

Answered

Comments

6 comments

  • Avatar
    Wishgranter
    Hi Fredy

    Im see the screenshot but see there just the Java script issue ( win 7 related )
    0
    Comment actions Permalink
  • Avatar
    fredy
    Oh, my bad!

    Fixed the link: https://cdn.pbrd.co/images/2vs0G3Q.jpg
    0
    Comment actions Permalink
  • Avatar
    fredy
    I am getting radically different results from the same set of photos and with the same workflow/parameters in RC.

    Why is it randomizing the output? Below is the result of a third or fourth attempt on the early shot.

    http://pasteboard.co/3eDi301.jpg
    0
    Comment actions Permalink
  • Avatar
    Wishgranter
    Hi Fredy

    Are you using fixed camera scanner ?

    If use a fixed camera setup then it would be better to use XMP workflow, a precalibration so you use a good working alignment on different scanns.

    Need same file names and not changing camera positions between scanns
    Align and inspect good scan = "Project A"
    select all cameras in ALIGNMENT ribbon
    hit XMP tool and export as LOCKED = the XMP are saved next to images used in project
    look in folder with the images for XMP files = select all and COPY

    move to second folder "Project B" with images ( the one what need align )
    PASTE the XMPs

    load images from "Project B" to the RC
    align the data ( XMPs are used automatically so no need to set anything )
    Should get good alignment

    the XMPs transfer all parameters describing images, and just detect actual subject inside

    can hit RECON
    0
    Comment actions Permalink
  • Avatar
    martinb
    Dear Fredy

    I am getting radically different results from the same set of photos and with the same workflow/parameters in RC.


    This is a strong indicator that your cameras are in setup where you are dancing on the edge of some precision thresholds.
    Inspection tool is the one of the tools which might help you to find a more optimal configuration for your cameras.

    Another option is to register cameras with "a well conditioned scene" such as boxes covered with news paper. Then you can export camera poses using XMP files and recycle camera poses like Milos proposes.

    You can also group camera calibration parameters - say, constraining a set of cameras to have the same focal length. This makes cameras to sit better in such situations. You can group camera params automatically by clicking on images root in 1ds -

    group-params.JPG


    or you can make RC to group images automatically when images are add - change application settings \ Import settings \ Group calibration by exif to true

    or you can group images individually in 1ds - select one or more images and set the same group number in "Calibration group" or "Lens group" fields. Use -1 if you do not want to group parameters.
    0
    Comment actions Permalink
  • Avatar
    fredy
    Hey guys,

    Thanks a lot for your guidance.

    I´m trying to make a model from our new raspberry pi scanner, to use as template project and to put the XMP workflow to test, without success yet.

    I realized that I´m getting split parts on different components. If I select lets say, component 1, I can generate another mesh, but with missing parts and diferent rotation on the scene. I wonder why that is happening and how to fix it.

    Below there are links for some screen shots. Also, what does that exclamation icon mean?

    Thanks a lot, again.

    https://cdn.pbrd.co/images/cobCla9.jpg

    https://cdn.pbrd.co/images/cocZaAh.jpg

    https://cdn.pbrd.co/images/coexbxZ.jpg

    http://pasteboard.co/cogfK5R.jpg ( Left: high on agisoft / Right: normal on RC)

    By the way, I didn´t define the ground plane nor set region because I wanted to have the most direct raw result to see what would happen. All models were done in normal quality.
    0
    Comment actions Permalink

Please sign in to leave a comment.