ok so this has been driving me nuts for like two weeks. doing a survey flight over a limestone quarry, M300 with the P1, ran the mission through DroneDeploy at 120m AGL with 75/65 overlap front and side. processed it in Pix4D and the orthomosaic has these weird smearing artifacts right around the active extraction face ? like the wall of the quarry is just kind of melted looking and the pointcloud in that area is a mess.

my GCPs are all on the upper bench so maybe thats part of it but i wasnt expecting it to be this bad. the flat areas of the quarry floor look totally fine. is this a known thing with steep faces or did i mess up my flight plan somehow. thinking maybe i need to do an oblique pass or something but not sure how to set that up in a way that actually helps the processing

yeah this is a classic problem with vertical or near-vertical surfaces, photogrammetry really struggles with them from a nadir-only flight. the algorithm doesnt have enough parallax to reconstruct steep faces properly so you get that melted look. oblique passes help a lot ? some people do a dedicated orbit around the face at like 45 degrees and merge the datasets. its more work in processing but the results are way better.

also worth checking if the quarry face is causing shadowing at your flight time, harsh shadows on textured surfaces can really confuse the feature matching. if you can fly when the sun angle is more frontal to the face that helps too. your GCP placement is probably a factor as well, ideally youd want some down on the floor level too but i get that that can be a pain in an active quarry.

had almost the exact same issue on a open pit mine last year. ended up doing a double grid at 60 degrees crosshatch plus a manual orbit at 40 degree gimbal tilt around the highwall. processing time was brutal but the face came out clean. Pix4D handles the merged datasets ok as long as you set the coordinate system consistently before you import everything, learned that the hard way