I reckon STMANI may be a little difficult to install, in part because it is a Linux program. Even for Linux experts it is not free of issues, as JackDesBwa can attest. I want to thank him for his helping comments, that allow me to improve this.
Sure the old version had dependencies difficult to satisfy today. The dependencies of new version are easier to fulfill and actually they installed without problem on my system. On systems based on Ubuntu (except likely the newest one), it is a bit less straightforward but not hard (there is a naming convention different than other distributions that cause problems but can easily be fixed thanks to the notes in the installation guide).
However, I am looking forward to the english manual to be able to use the program for real.
If you remember my statement that "Anything can be aligned", STMANI3 is the program that I use to prove... at least some of the test cases:-)
I was not on this mailing-list when this discussion took place, but the statement is quite obvious when you do the math. However, it comes with some limitations on the field of view for example, since you can only have stereoscopic effect in the zone seen by both cameras. Even if I read only parts of this discussion, I was happy to see confirmations of what I saw in the equations too.
In the last days, I drew an illustration of this with voluntarily exaggerated divergence (left/right and up/down) as well as a rotation of the cameras around the optical axis. It was not motivated by this thread, but it might be an interesting addition to the topic: https://stereopix.net/photo:koUNdrW2kF/