Just reading about Orthophotography. Articles point out that usually orthophotos are only succesful, by simply modifying a photo, for nearly-flat surfaces incl building elevations that have little in the way of projections and recesses. (Is this what used to be called rising-front photography?)
Is that why people e.g. Gotz use photogrammetry, to first create a super-photorealitic non-perspective 3D model from which a 2D ortho view can be taken?
Is that right? Not sure I still don’t quite get it.
Hi Tom, yes, that is the reason! 
The “old” method is, as you say, only suitable for flat surfaces. For everything else, there will be a noticeable perspective in lines that are not parallel to the camera sensor. The result will look like it’s been taken with a tilt-shift lens.
I would argue that the “new” photogrammetry is at least as quick, possibly faster, since the “old method” requires a minimum of 5 theodolite measured points per image to be metrologically accurate, provided you have a nearly undistorted image. And then there are the inevitable color differences from one image to the next, which the “new” method takes care of automatically…
V gd - now I have a name, and there’s a developed workflow out there, for part of what I have in mind.