Another photogrammetry based PBR texture done to practice and improve my current workflow. Because I have found moving on sand might be quite destructive for surface itself, I decided to change my usual way of capture and experiment a bit more with different shooting paths to see the difference.
In that capture I took series of images from outside and when all sides were covered I moved towards center to capture central part.
In result I captured 31 images and used them to build 25mln highpoly model in Photoscan. Next I built lowpoly in Zbrush and baked diffuse, normal and ambient occlusion maps in Substance Designer. I tiled them and patched manually in Substance Painter and processed again by Designer to fix color and normal inconsistency across the tiles and generate height and roughness maps.
Shooting path conclusion: might work if area is well defined and measured. Otherwise.. there no turn back if it was too small. Amount of captured data is better than usual row by row.