okay so this has been bugging me for a while and i cant figure out if its a flight planning issue or something in my processing settings. im doing a construction site survey, pretty standard stuff, and the point cloud looks great over most of the area but on the northern edge of the site its just noisy and kind of blurry, the mesh looks awful there and the orthomosaic has some artifacts too.

the thing is the GCPs are spread across the whole site including that corner so its not a control point issue as far as i can tell. flying double grid at 75 overlap both directions, 80m AGL. processing in Pix4Dmapper. the site has some tall equipment on that edge, like a concrete batch plant and some silos, but ive mapped around machinery before without this kind of issue.

anyone have ideas? im kind of at a loss here and the client wants an updated deliverable by end of week

the silos are almost certainly your problem. reflective or cylindrical surfaces absolutely destroy photogrammetry matching in that area because the feature matching algorithm cant find consistent keypoints across images ? the surface looks different from every angle and theres probably shadow variation across the day too if you were flying late. you basically cant get clean reconstruction on smooth metal cylinders with standard RGB cameras, some people try to work around it by masking those areas out in the processing but then you just have a hole in your model instead of garbage geometry which is arguably better for a deliverable

also worth checking what time of day you were flying on that edge ? if the sun was low and casting long shadows from the silos across your ground features that can really mess with the matching on adjacent areas, not just directly under the equipment. i had something similar on a warehouse site and just reflying that section at a different time of day around solar noon fixed most of it. not sure if thats an option before your deadline tho