okay so this has been bugging me for three days now and i cant figure out what went wrong. did a surveying job over a construction site, laid out 8 GCPs with a rental RTK GPS unit (Emlid Reach RS2), shot the whole thing with my Phantom 4 RTK on a double grid at like 80% overlap both ways, processed in Pix4D and the orthomosaic is just... warped. edges are bowed out and there's a really obvious doming effect in the DSM which is exactly what GCPs are supposed to fix.

the GCP coordinates look right when i check them against the survey points the client gave me, i marked them in Pix4D the way ive always done it, initial processing looked normal. im just stuck. could it be the RS2 losing fix during acquisition and giving me bad points without me realizing? or is there something in Pix4D im overlooking in the calibration settings

doming is almost always a GCP distribution problem in my experience. 8 points sounds like a lot but if most of them ended up clustered toward the center or along one axis the edges will still dome. you really need them spread to the perimeter corners of the site, like physically as close to the edges as you can get them, with maybe one or two in the middle. can you share a screenshot of where they were placed on the map?

also worth checking the RMS errors per GCP in the quality report, if one of them has a Z error that's way higher than the others thats your culprit and you can try checking it as a checkpoint instead to see how it affects things

had almost the exact same issue last year. turned out my RS2 had been in float not fix for like half the GCP measurements and i didnt catch it because i wasnt paying close enough attention to the status on the controller. the coordinates looked plausible so i didnt question them. started being really paranoid about confirming fix lock before every measurement and havent had the problem since. not saying thats definitely your issue but worth going back and checking your RS2 logs if you saved them