I was hoping someone could help me improve the quality of my 3d model. I’m very new to DD (only 3 maps done so far), so please bear with me.
I mapped the same location twice. Once in structure mode and once in terrain mode.
- 500 images
- 276 Nadir (35m altitude, 80/80 front and side overlap, 24 speed)
- 224 Oblique (orbit flights around each structure. 35m altitude, 50-70m radius, about 40-50 pictures per structure)
- Captured 2pm, good lighting.
- The goal of this run was having a good 3d model.
- 375 images
- All nadir (35m altitude, 80/80 front and side overlap, 24 speed)
- Captured 4pm, good lighting.
- The goal of this run was generating a good Ortho.
My goal is to have as much structural detail as possible, hence why I chose structure for the 3d model oriented run. However, I was surprised to find that Terrain processing produced a comparable model to structure, despite using less pictures and having no oblique images. In fact, it seems that terrain actually edged out structure in terms of rooftop detail.
So my question is: what are the ideal settings/best practices for mapping urbanized terrain? The vast majority of my work will be mapping and modeling for in progress job-sites, so structural detail is very valuable to me.
Would I be better off using terrain, where I have a 3,000 image limit and just add cross-hatching and many more oblique layers?
Adding this here since I can only post 2 links as a new member:
Here is another example of a map processing with “terrain”. http://drdp.ly/wrafae
In this case I got very good details on one of the rooftop AC units. I did however, fly much closer for my obliques. I was probably on 20m radius vs 50-70m.
I took a look at your maps and the number of oblique images does exceed what I would generally recommend within a map of this size. I would recommend lowering your altitude when capturing oblique imagery and increasing your altitude when flying the normal mission (Nadir only).
I count almost 8 orbits on this mission once you had finished the initial Terrain portion of the mission. I would recommend something more along the lines of 3-4 orbits with much less overlap between the images being captured. I would recommend flying them manually and only capturing as much as needed. The problem with oblique imagery is that the number of obliques/altitude/angle of the camera will vary based on the area of interest being mapped. This might take a bit of experimenting to get the correct data for a given site. One thing to remember is that you can have too many obliques.
Depending on the area of interest you will sometimes capture the sides of building much more efficiently by just using the terrain option and increasing your front and sidelap. The “cross-hatch” option for flight planning is something that is on our product roadmap as a potential feature to integrate into the flight app.
3D mapping is difficult and can take quite a bit of practice. The more you fly an area of interest the more comfortable you will become with choseing settings that will work best with the structure being modeled!
Please let me know if you have any questions moving forward.
Thank you Zach! This is super helpful.
I’ll give taking less pictures a shot and try and decrease the overlap of the orbit sections.
If I could add one more question - at what distance should I be taking my oblique shots? My current radii stands at anywhere between 50-70m. Would it make sense to cut that in half? How close is too close?