Two directions, and they solve different problems
The first is one source face onto many target photos — putting the same person into a set of twenty shots. That is the common case for a photo set from one session, or for producing variations of the same concept.
The second is many source faces onto one target photo, which is how you test which source image actually produces the best result. Instead of guessing which of your photos makes a good source, run several against the same target and compare the outputs side by side.
That second direction is worth knowing about, because source choice matters more than most people expect: angle, resolution and lighting in the source drive the quality of the swap far more than any setting does.