A pawn can unmask an attack even when starting from its original position. Notice that at the start of a game each side already has the kernel of a pawn discovery in place: the pawns on d2/d7 and e2/e7 mask the bishops behind them (the queen is poised to be unveiled, too, but set it aside for now). If Black puts a piece near the middle of the board and then leaves another on one of the bishop’s diagonals, he may be asking for trouble. Here White can unmask an attack on Black’s queen by his bishop with d2-d4, also attacking and winning Black’s bishop after the queen moves.
By the way, notice that Black almost can win the exchange here. Suppose he replies to d2-d4 with Qg6. Then after White plays d4xB, Black still has the pawn on g2 pinned. He plays Bh3, threatening to mate with Qxg2. To stop this, White can play g2-g3—but then Black wins the exchange with BxR. What prevents this is White’s queen on d1, which he can move to f3 instead of playing g2-g3. It's a pattern worth knowing.