I use a canon S100, CHDK and KAPscript.lua for mapping. The S100 clock is GPS synced, and the KAP script outputs a log file with the shutter release time to 10ms precision, including the image filename.
Is there any way to use the (possibly) more accurate shutter time from the KAP log instead of the exif data from the images (which I assume is what the MP geotag utility uses)? Would be great, particularly in combination with an RTK system on GPS 2…
Happy to provide images, log files, or anything else that might help.
I did have some time to investigate the possibilities of geotagging using the KAP log and an Emlid Reach RTK system - I was able to use the KAP log time stamps to geotag ~500 images from a mapping flight and using Pix4d to verify the image locations, ~90% were within 15cm of the tag location, flying at 7 m/s, which demonstrates the accuracy of the timestamp in the KAP log file.
However the GPS clock sync is not perfect, to get that accuracy I had to offset slightly less than the 17 second GPST - UTC offset, 16.318 seconds in this case. Have not had a chance to see how that varies, but I suspect the camera internal clock is only syncing to the nearest second.
Still, using the KAP log and trial and error to find the offset, and postprocessing the GPS logs with RTKpost gave pretty good results. With 14 check points surveyed the average localisation error was ~5cm x axis, ~1mm y axis and ~33cm z axis, sigma and RMS error low as well.
The same image set tagged using the mission planner utility and the copter flight logs with the standard GPS gave geolocation errors of metres.
Unfortunately the mavlink connection providing corrections to the rover unit on the drone is not stable enough to guarantee that level of accuracy throughout the flight, still working on that…