Agisoft Metashape
Agisoft Metashape => Bug Reports => Topic started by: gantar1 on May 10, 2021, 10:04:31 AM
-
Hey,
i'm traying to build ortomosaic with photos of DJI MINI 2 with ground points. Result with option rolling shutter enabled is worse than if rolling shutter is disabled. Red and cyan lines are correct. What can be wrong?
Best regards,
Kristjan
-
Hi gantar1,
to check the accuracy of the two alignments, with or without rolling shutter, I would use some GCPs set as check points during optmizing process (assuming tou have at least 3 to use as control). This way you will get an independant assessment of accuracy....
Comparing roof footprints against ortho will not do it especially if you did nor use a precise DSM for rectification (true ortho vs just DTM based ortho where roof outlines are not accurate)....
-
Thanks for reply, we have controls measured with GPS (Leica 1cm). with rolling shutter off it looks OK. In both cases are heights wrong (cca 1m difference). This problem is only on mini 2. With Phantom 4 V2 is everything OK.
Kristjan
-
Kristjan,
maybe you can show calibration page from report to see if there excessive are values for adjusted cx, cy relative sensor size or if f varies greatly from initial value (around 2785 pix). We have experienced some similar problems with DJI Mavc 2 Pro....
-
Hey,
here is report.
-
Hi gantar1,
yes it seems the estimated value for f is quite off the starting value (4490 µm / 1.76 µm/pix = 2550 pix) and cy is about 33 % of half sensor height. This happens for Mavic 2 pro when flying with camera always pointing in same direcion regardless of turns. For example for N/S lines drone will first go forward to North in first line then backwards to South in 2nd line, etc...so top of camera will always face North...
I would try to do a new alignment wth f, cx, cy, b1 and b2 fixed to see if you get better results...