Right. Such as, extras in movies. Because, "350 million public faces" does not seem "cognitively digestible", but it could just be «350 million images» including large variations of the same faces, consistently with the "frames of video" idea.