What happens here is that when you create a DEM, the number of rows and columns is no longer as in the point cloud. Then, pc_align produces an aligned cloud, but it has the same dimensions as the DEM, rather than as the input point cloud.
So, point2dem cannot function with the L.tif that is 1-to-1 with the PC.tif before being gridded with point2dem.
I guess what you can do is:
- Use pc_align with the input PC.tif, before gridding with point2dem, then the output of pc_align will have the same dimensions.
Or:
- Create the orthoimage with point2dem with the original cloud, producing a DRG, then use pc_align as now, get a point cloud which would have the same dimensions as the DEM (I hope), which has the same dimensions as the DRG, then call point2dem with this DRG instead of L.tif.
I did not test this, but given that the problem is with mismatch in size, it should work. I don't recall now if point2dem will be happy enough to accept the DRG.tif instead of L.tif if called with the --orthoimage run/run-DRG.tif option, or if one has to first rename DRG.tif to L.tif.