Film's too big. Make it smaller. Start your canon.
In Harold Bloom's Canon of Western Literature, Bloom proposes beginning a canon by first selecting the one author without whom a canon cannot reasonably exist. Then you pick the one work of that author that is absolutely essential. There's your logical first work in the canon. Decide why that work deserves to be canonized--what qualities it has--and use that information to develop criteria for including more works. Bloom's suggestion: the higher the bar is set, the better the canon will be.
A canon is not necessarily a list of favorites or democratically selected best movies, although it can be. A canon is simply a selection of works that are somehow essential to their medium. The purpose of the canon is to winnow a huge number of works down to the ones most deserving of discussion. I recommend steering clear of nominating your own favorite director and film, and instead taking more of a historical/developmental approach. Just to shake things up.
Since it's much easier to pin down the author of a piece of literature than the author of a film, we can go with the theory that the director is the author of the film, just for argument's sake. This was the method Paul Schrader proposed in his abandoned film canon project.
Looking at the entire sweep of movies from square one up to today, I would say the filmmaker that had the most instrumental effect in defining film is probably Alfred Hitchcock. Hitchcock learned his lessons from the past and left lessons of his own for the future. He used the classical language of film more impeccably and consistently than just about any filmmaker I can think of. He understood the effects of images and sound on the human psyche as well as anyone ever has.
I'm not saying he's the world's greatest director or the earliest director to do some of the stuff he did, but he's the one who, on the whole, carved the biggest mark into cinematic history. So he's my suggestion for the "author" without whom a canon of film cannot exist.
As for which of his works is most essential, I'm actually not quite sure which I'd pick. It would have to be the first major work of his where everything came together tightly in that Hitchcockian way. I'm still trying to decide which one that is. Strangers On A Train might be it.