Modelling underwater caves from video

Discussion in '3D Design Requests' started by Liz_Rogers, Nov 11, 2012.

  1. Liz_Rogers
    Liz_Rogers New Member
    Hi all,

    I'm a cave diver (yes, scuba diving in caves that are full of water) interested in 3D modelling. At the moment mapping underwater caves is done using a bit of string with knots in it, a depth gauge and a compass. We've had some success with creating 3D models of individual underwater chambers using overlapping photographs. The restriction is the number of photos required and the size of the cave (8km/5mi of interconnecting underwater tunnels).

    So....is there anyone who can help me out with developing a dual video to 3D map technology? We have invested in a dual Contour camera set up and have taken a bunch of paired video footage from the caves. I've seen people creating models from dual video output on land, but only prototypes and nothing commercially available as yet.

    Alternatively, is there a way to create each section of cave from photos, and then stitch the sections together?

    For those who are interested, you can check out my underwater cave photography to get an idea of the environment we're trying to model.

    Thanks in advance for any and all leads or assistance!

    Liz Rogers
    http://lizrogersphotography.com
     
  2. stonysmith
    stonysmith Well-Known Member Moderator
    Have you tried 123dcatch
    This sounds like a cool project.
     
  3. Liz_Rogers
    Liz_Rogers New Member
    I think it's cool. It will revolutionise underwater mapping if we can get it up and working reliably. And I hate counting knots in string.

    Yes, the success so far has been with 123catch. See here for more details: http://lizrogersphotography.com/2012/10/3d-mapping-the-pilla r-in-tank-cave/ The link to the model generated is halfway down the page.

    I'm not sure that it would work so well if I tried to insert 5,000 photos covering 5 miles of cave, however!
     
  4. stonysmith
    stonysmith Well-Known Member Moderator
    Option #2.. did you try Microsoft Photosynth? http://photosynth.net
    It may be capable of handling those 5000 photos.
    At one time there was an option to get a point cloud out of it.. but I'm not sure if the option is still there or not.