okay so this has been driving me insane for the last two weeks. im doing a survey job, decent sized site, using 6 GCPs measured with a pretty good RTK setup. running everything through Pix4D mapper. the orthomosaic looks fine visually but when i go to check my GCP residuals in the report some of them are way off ? like 8-10cm horizontal error on a couple points and i cant figure out if its a collection problem, a processing problem, or if something is wrong with how im placing the GCPs in the software.

flight was at 100m AGL, 75/75 overlap, about 400 images total. GCPs are spread across the site, not just around the perimeter. i thought that was supposed to be good practice. the weird part is the points near the center of the site are fine but the ones toward the edges are the ones throwing high residuals. anyone dealt with this before?

edge GCP errors like that usually point to one of two things in my experience ? either the image marking in Pix4D is off (like you marked the wrong pixel on a couple images, easy to do if your GCP targets are small or the images are a bit blurry near the edges due to rolling shutter) or your GCP coordinates themselves have some issue. what size targets are you using? anything smaller than about 40cm tends to get really annoying to mark accurately at 100m altitude. also worth checking if your camera calibration is getting weird ? sometimes if you have a lot of nadir-only images and not much oblique overlap the model can dome slightly which shows up worst at the edges.

might be a dumb question but are you using checkpoints vs control points correctly in pix4d? i made the mistake early on of marking all my GCPs as control points and not keeping any as independent checkpoints, so i had no way to actually validate accuracy. the software can kind of overfit to the control points and the residuals look good in the report but the actual accuracy is garbage. just something to double check if you havent already