In my lab on images formed by converging lenses (measuring image and
object distances and focal length etc...) I have them set up a real
image on a screen and then slowly cover the lens with an opaque card (
from the top, bottom, left right sides of the lens) and then explain
what they see. Those that understand why there is always a complete
image present have made the leap away from just the principle rays and
those who haven't taken this step are forced to confront it as they
write their explanation.