Drift in hort / vertical accuracy from photo 1 to 800. Bad processing results


We have a 800 image project with a M300 RTK. Emlid RS2+ Base station established with NTRIP FIX. We’re finding some large errors between our processing RTK data processing vs the checkpoints. Really scratching our heads to why but have noticed huge drift in the accuracy of the images.

The first images have 0.038m Horizontal accuracy and 0.100m vertical accuracy.
The last images in the same mission drift by up to 0.156m horizontal accuracy, and 0.454m vertical.
Is that normal for the accuracy to change throughout a mission? Any idea to why that is happening? Is this why our outputs are so off, as verified via our check-points?

Thanks for any help.

Hey Scout,

We are experiencing the exact same issue with our M300 RTK, Emlid RS2+ and NTRIP from CAN-Net. Really scratching our heads trying to figure out the issue but can’t seem to figure out why. It doesn’t happen on all flights.

We flew one mission with a base station location determined with NTRIP average FIX (2min) which had the errors but the exact same mission flown just after with a base station location determined by Average single (20min) did not have the issue.

Hoping Pix4D can provide a reason.

Hi guys,
From our perspective, PIX4Dmapper simply reads the accuracy values that are written to the images. So, this is an issue with the drone and/or base station.

If you are seeing errors with your checkpoints, then please attach the quality report to this posting and we will take a look.

We also have 3400 images project with a M300 RTK. While importing images into PIX4D, I noticed huge drift in the accuracy of the images in an alternate basis. Images were taken with multispectral sensor that has six band. I noticed that 1 set of images has 0.012m horizontal accuracy and 0.024 m vertical accuracy while next set has 5.00m horizontal and 739.00 vertical accuracy. I have noticed this problem only in some flight. So, I was wondering is this the problem related with Pix4D or with sensor or drone. Any idea why is this happening?

Mapper is simply reading the accuracy from the image Exif data. I would check with the manufacturer.