Has anyone been able to test out the new RTK and see if it can produce better models? I have a problem with many of the models that drones generate of complex buildings (I work in historic buildings, where there are lots of fiddly bits, as well as ruined walls, no straight lines etc.) which can look very smoothed - as if they have been made of plasticine or play dough! Even surveys with the most detailed oblique images tend to produce this effect, which I suspect is a product of the stitching and the wire frame generated.With more precise located images, could this be improved? It would be really great if someone out there could do a model with and without RTK ti see if it makes a difference! Thanks! Mark
Definitely something I want to see. I doubt there will be much of a marked improvement because the 3D model represents the images draped over the mash which means that if the mesh isn’t tight enough then the image gets wrapped unevenly. Creating an accurate 3D mesh depends more on 360deg capture than the absolute accurate position of the drone. From what I am hearing you describe those details sound very hard to capture with photogrammetry. It would either be a very extensive oblique flight plan on multiple levels or a laser scanner.
I’m product manager for flight and processing here at DroneDeploy.
We’re working on improving our 3D reconstruction code all the time, and it sounds like you have some good examples that we can test with. Crisper 3D models and edges are certainly possible. Could you share the link to an example of a map where you had poor results despite good input imagery?