I am looking for suggestions to solve an thermal map issue. The flight was for a commercial building inspection. Flight was at 42 Meters with 82,82 overlap. (this was a flight setting that I changed. Usually use 83,83). Zenmuse XT Camera 640 radiometric, 13mm. Flight pattern was east-west.
For processing I chose default settings and only change radiometric source to camera. (as I always do). The issue in the rendered map is the south building roof edge is very soft.
I tried to re run with PC set to high. That attempt returned a fail to process result.
My question is if anyone can suggest what setting to try to sharpen the south edge. I looked at the single images and they look good, exact quality as the north roof edge which presents perfectly in the map.
BTW, I pretty much use WebODM for all of my thermal map rendering. My projects with 83,83 overlap are usually between 80 and 140 photos. This map rendered in just under 10 minutes. All I can say is I am very grateful.
Only ask I would have is:
The ability to measure locations from within WebODM without having to go to QGIS and using the measuring tool. We already have some measuring tools in both 2D and 3D. Flir tools has nice spot, box (min, max) and line measurements for single photos. I think that is an excellent set. We already have great palette options in ODM.
When we select “export” from the Layers tool box in 2D view, it would be great to somehow include a legend based on how we have adjusted our min and max values. If we are only interested in relative temps, it is fine as is however a legend would make the image produced way more meaningful.
If nothing changes and ODM continues to provide this excellent rendering tool, I will also be very grateful. I am useless when it comes to coding so the best I could do is offer to contribute to a funding effort and/or contribute data sets.
I would suggest perhaps implementing a secondary “orbit” survey with the gimbal at at few different angles from nadir (15, 30, 45, 60) or so if you want to get really crisp building edges.
Without seeing all the input data, it would suggest that something on the edge made it less feature-full and easy to get tiepoints from. Could be excessive shadow, too bright, or something else.
Yeah, I’d say something similar and only add the the edge of your building is very close to the edge of your flight area, so the overlap is likely much lower in that part of the flight plan than it is away from the edges.
Maybe I should not have included the 3d image. I do not care at all about that result. I only included it to demonstrate the reconstruction. It is only the Ortho that I am concerned with.This is the first time I have had an issue like this. I just looked and the images are very good (same quality as those providing info for the north edge of roof) and include two passes of the drone which included full view of that south edge of the building with that 82, 82 overlap, also the same as the north edge and normal to what I always use as a flight pattern. I have tried to render with different settings and get the same results. This one is a mystery to me.
I am going to chalk it up to my possibly pushing the south borders of the flight plan slightly too far north. I wanted to be sure that I did not fly over people. Store entry is on south side.The report looks like the camera may not have fired in a couple of locations on that path.
I guess it is important to own a fail. I clicked on camera locations and I was surprised to see how far north I did push the first track of the mission. There was people entering and leaving the store that I did not want to fly over. I will remember to display camera locations in the future if I have results like this. When I was watching my drone screen, it did not look like I was that far from the edge.
