Process vs. Thread for multithreading

[mprat]

The example in multithreading_opencv.py uses Thread objects rather than multiprocessing.Process objects in Python.

I have an implementation that simultaneously is attempting to stream from 2 cameras in separate Process objects rather than Thread objects on an Ubuntu machine, and we are seeing some inconsistencies in how Vimba behaves. We have two GigE cameras connected on separate NICs.

Is there a reason that Threads are preferred? Is there an issue fork-ing the Vimba singleton in Process objects that could be causing an issue? The default behavior in creating Process objects on Linux is fork rather than spawn, so it is possible that using spawn does not create a problem for the Vimba singleton?