Learning - Multiple Altitude Layers - Construction Site

Answered

Comments

6 comments

  • Avatar
    Dillon from MCWB Architects

    If there is suitable overlap, and your images are of similar exposure/lighting/etc, the default alignment settings should be enough. If they aren't, I will occasionally toggle camera priors true/false in the settings and find this helps with alignments in which image overlap is good. You might also try turning on "Background Feature Detection" in alignment settings, if the majority of the photo is in focus. 

    Finally, if you are getting good results, just in multiple components, you can play with the "Force component rematch" setting, and the "Merge components only" setting in the alignment tab.

    What kind of drone are you flying, and what kind of overlap do you have in your settings?

    0
    Comment actions Permalink
  • Avatar
    Erik Kubiňan CR

    Hello Mr. Johnston,

    It is very unclear what you have said so you could possibly explain better but, to make it a bit quicker for you a model should never be shot in just a loop of images you should be having more elevations with various angles basically instead of a circular line around the model do a grid looking pattern to cover much more and define the shape in 3D space by looking at an object from different angles. Side 3/4 frontal 3/4 side.. etc..

    0
    Comment actions Permalink
  • Avatar
    Douglas Johnston

    I’m performing either a row grid or cross grid for construction images. Using Maps Made Easy - Map Pilot to collect the images, at 280 ft altitude, Inspire 2 with a X5S & 12mm Zuiko lens. Map images collected at 70, 70 or 80, 80 overlap. 

    The 2D Photogrammetry normally looks nice, When either processed by MME or Metashape. The 3D looks poor, regarding the structure.  If I perform an additional loop around structure at 100 ft, altitude, gimbal tilt at 70%, overlap 80,80. And process individually, Separate from video grid collection, I get a nice 3D image. When processed all together, bad 3D of structures. 

    0
    Comment actions Permalink
  • Avatar
    Erik Kubiňan CR

    Well okay now I've just understood what's going on. I would not recommend you using video for reconstruction where you seek any good detail, it is possible to work with a video in RC but normal shooting is much better. Normally when shooting a video the quality is much lesser than the quality and detail in shots obviously and also, a motion blur can be very present in a video footage, therefore please prevent using video while expecting a high detail reconstruction. 

    0
    Comment actions Permalink
  • Avatar
    Douglas Johnston

    I'm not using video, My statement of "video grid" was not clear... I meant the image grid and the image structure loop.. all images. 1 set is the higher grid and 2nd set is a loop around structures.

    0
    Comment actions Permalink
  • Avatar
    Erik Kubiňan CR

    Okay Sir at this point it is quite unclear on what exactly is going on and where the issue could be. Although I do want to help you of course. Please write an email to support@capturingreality.com and provide us with as clear as possible description of your issue, how you shot it, what does the hardware for shooting look like and some pictures from the RC application (component, count of pictures, few pictures and the viewport from different angles with cameras enabled and then finally the resulting mesh with both the good and bad quality as mentioned above)

    0
    Comment actions Permalink

Please sign in to leave a comment.