Thanks for the details! This technique works surprisingly well. (One of the artifacts I was able to find is at the base of the Bay Bridge in SF's SOMA -- there's a vertical wall that has the street surface projected up along it rather than an actual hole underneath the bridge. That does seem like a challenging case for airborne lidar + stitching.)

Another impressive spot is the top of the Stratosphere Tower in Las Vegas -- it manages to capture the spike at the top fairly well. It'd be interesting to know how much hand-editing they did for sites of interest like that, and how they represent hand-edits in a way that can be re-applied when new lidar datasets come in.