Hi, so the data i have is approx 300 photos with a Nikon D810 (with or without geo -location data sideloaded from a phone app) and a .e57 file created from multiple scans with a Leica RC360. of a building.
No matter what I do I cant get to the reconstruction phase without the error as stated in the title "Positioning levels of photos and point clouds are not consistent "
If i try to manually set the location of the scan as 0,0,0 it gives an error "More than 1 scan in E57 file"
So I assume there must be a setting for the software to align the point cloud and photos visually and not their geodata?
The geolocation of the model is not required. Just need the model as 'proof of concept'.
The only other point to mention is that asked our 'scanner guy' are co-ordinate systems for the Leica and he wasn't aware of any settings for projections.
Are you sure that geolocations from phone are actually accurate?
For pointclouds to work they need to be in one coordinate system so you will need to read some GCP from point cloud and use them to align photos and only then you can add point clouds in processing.
The Nikon photos are not particularly accurate.. You sync the time on the camera and phone and then back on the desktop the app sideloads the location to the photo based on the phones location at the time the photo was taken. Best case scenario is standard GPS accuracy of 5m but I would assume up to 10m.
So there is no method for a purely visual alignment of the point cloud and photos???
Of course it will not align automatically you can't expect much accuracy from phone GPS. Vertical accuracy is non-existent which you can check for yourself after aerotriangulation the altitude will be +/- 20m which is too much for automatic alignment with point clouds. They should be within meter or less the same coordinates.
As stated in my first response I expected low geolocation accuracy, but georeferenced accuracy was not the object of this project. It was purely to test workflows for reality mesh creation. I am doing trials with various software packages, and Realitycapture has merged the laser scan and the images without any geolocation data. As stated above I will try again with control points.
Yes it seems that other softwares are catching up in development and features. Honestly the alignment with point clouds in Contextcapture is not explained in details so not sure what to expect in different situations.
Yeah GIS is my main job, but I also do drone photogrammetry and with the same photos in Pix4D, Agisoft Metashape and Drone deploy gave different results with 3D models and orthophotos! I think you need to test and find the best software for each task!
Contextcapture gives better accuracy if all goes well and it is also gives you models which can be directly used in Bentley software. But for orthophotos it takes too much processing.
Hi Oto,
For hybrid processing, do you use colorized scans or just black&white?
I just encountered an issue that the colorized scans tended to automatically compensate the place with a random color where the scanner couldn't reach.