Does it stop at the divide by zero, or keep going? There are a few places where it is computing band statistics on a per grid cell basis and I think it just runs in a naive mode so that if it hits a 'bin' with no points it still executes the command (like mean, or SD) which with no or one point would result in a divide by zero.
It stops at the divide by zero.
I am working with the Curtis group at Ohio State and this is one of our last steps. Do you happen to have any further recommendations?
Is this using the main post-processing module from the GUI where you just supply the LOG file and OUT file (and optional dense PLY file?) You said it seems like it is crashing on the telemetry conversion step, could you paste in some of the console output leading up to the Divide by Zero error?
The telemetry data is my only guess as to why it is crashing. We received the OUT file from Photoscan pro, so I would assume that is correct.
I have attached a screenshot of both the error and then what I am inputting into the GUI. Hopefully this is enough information.
OK, yes, it looks like it is totally failing in the telemetry stage. you are getting a divide by zero because the variable GPSlength must be empty or zero, thus causing the whole thing to crash. Unfortunately it looks like we are running up against a specific challenge with the way EcosynthAerial was designed. We wrote the program to automatically parse the ardupilot LOG file, which is what it is requesting in the post-processing GUI. It appears you are supplying it with an XYZ textfile, which does not have the exact same formatting. I think it would be possible to change the code to check for such a thing and to have it use one format or another, but we don't have any active developers at the moment. I can try to fit it into my schedule, but it could take me some time to get to it. You might try two things: supplying the LOG file directly - or following this set of instructions http://ecosynth.org/profiles/blogs/full-campus-scan-with-octo?xg_so... from Stephen and using the attached scripts to generate a Photoscan georeferencing file and then produce outputs directly from within Photoscan. Note that you would need to skip step 3 in Stephen's instructions and use the parse_ARDU_log.py script. Sorry for this confusion.
I apologize for the late response.
So, what type of telemetry files does the post-processing within Ecosynth accept? I am afraid I do not have the LOG file, seeing as I was just given the trimmed telemetry data. Also, have you had any luck getting Ecosynth to accept xyz files?
Also, we have been having some trouble with the UAV shooting directly up into the air once we enter autopilot. It then goes well above the set height. Some have said that this may be due to z-axis vibrations. Any idea how to correct this or other recommendations? I have attached our Mission Planner script if necessary.