Support Website Contact Support Blog

Point Clouds Skewed on Z-Axis

My team has flown the same campus multiple times in an effort to create the most accurate point cloud possible but we are struggling to merge the projects effectively. 

Flight 1: 2 60m pre-planned grid flights at 70% route overlap.

Point cloud result: The point clouds came together well but we realized that we needed more coverage of the buildings on site.

 

Flight 2: 5 40m pre-planned grid flights at 70% route overlap, Terrestrial photos of the buildings (The drone was carried by my partner) with significant photo overlap, 2 40m pre-planned circular flights at 10% angle around the main building, and several manual flights on the corners of the main building to connect terrestrial to grid flights.

Point cloud result: Attempting to create a single project with all of these flights at once resulted in a point cloud with the buildings floating beneath the landscape, which was skewed at an angle of around -15%. After breaking this project into several smaller projects, the terrestrial photos seemed to come out ok but the landscape was still very skewed. Deciding to use the acceptable data from the first flights (60m, 70%overlap) for the landscape, we attempted to merge the landscape point cloud with the terrestrial point cloud using MTPs. This again resulted in the buildings floating in a void beneath the landscape.

Attached is a photo demonstrating the skewed landscape and floating buildings.

Is there a better way to connect grid flights with terrestrially taken photos?

Hi Jack,

If I understand correctly, the aerial photos are properly adjusted or located but the terrestrial ones are always skewed.

In order to merge different projects, they should fulfill several conditions:

  • The coordinate system (both horizontal and vertical) of the images has to be the same in all subprojects. Note: A subproject with no image geolocation can be merged with a subproject with image geolocation.
  • The GCPs have to be in the same coordinate system (both horizontal and vertical) in all subprojects. Note: A subproject with no GCPs can be merged with another subproject with GCPs. 
    It is highly recommended the GCPs to be added in the merged project, especially when more than 2 subprojects are merged.
  • The output coordinate system (both horizontal and vertical) of all subproject has to be the same. Note: A subproject with no georeference (arbitrary system) can be merged with a subproject with georeferencing.

In addition, the subprojects should be processed with the proper template, in your case, the aerial flights with 3D Maps and the oblique and terrestrial with 3D Model.

I would suggest to first process all of the aerial flights in one project, then the oblique flights in another project and then the terrestrial in a third project. I would then merge all of them as it is explained in this article

As the terrestrial project is the one giving troubles, I would especially check the coordinate system, scale, and orientation. In a terrestrial project, geotags are sometimes missing or considered the same.

I would also like to have a look at the quality report of the subprojects (especially the terrestrial one).

Regards.

1 Like

Thanks for the thorough response. Here’s the quality report for the terrestrial photos.

https://drive.google.com/file/d/1mcH-Tqe3NqVrtI06SGuyzYTYZp_icn1H/view?usp=sharing

 

Sorry if this isn’t the best way to share quality reports. I’m new to the forum

Actually, after taking another look, I realized I sent you a previous version of that project without all of the photos. I’ll update with a quality report of the actual terrestrial project soon.

Hi Daniel, I’m ready to post the quality report for the terrestrial flights but we would appreciate discretion when discussing this particular project from now on. Is there any way I could privately send you the quality report?

Hi Jack

Please open a support ticket:

  • Log in with your user and password

  • On the support website, go to “Contact Support” and complete the form.

Thank you very much.