Oh, I see,sort of like how the old school megaman camera works? It follows you, but keeps the next "room" hidden until you actually enter it.
Is that accurate?
I think the method I would use would be something like.
Have a camera sprite, the size of the viewport, it has collision turned on.
On a separate layer, I would build collision boxes to corral the camera box. This would allow the camera to be pinned to the player. The camera then follows, but is unable to move to far left/right/up/down when the player is next to walls or near a zone edge, due to the binding boxes you setup on a 0 opacity layer.
Does that make sense?