First off, 34 ♗xf6 removes the f6-knight defender of the <h7>-square, creating the threat of the standard mating sequence ♕h7+ ♔f8 ♕h8#.

Secondly, 34 ♗xf6 places the White dark-squared bishop on the a1-h8 diagonal, pinning the Black <g7>-pawn to this line. This pin prevents Black from meeting the mating threat by blocking the b1-h7 diagonal with his g-pawn (34 ... g6 35 ♕h8#). But the resulting position is hopeless hopeless and so Black resigns. He could have resigned after 34 ♗xf6 but probably played on because he was in time pressure.

Thirdly, the White bishop on f6 blockades the <f7>-pawn, so that Black cannot make luft for his king by playing 34 ... f7-f6. Since the support for the White queen arriving on h7 is provided by the f5-bishop and not by a pawn on g6 or a knight on g5 (both of which also cover f7), vacating the f7-square would allow the Black king to escape the backrank mate if not for the fact that the pawn on this square was blockaded by 34 ♗xf6.

And finally, 34 ♗xf6 puts the bishop on f6 where it controls the critical <e7>-square, the flight square that the Black king needs to escape his first rank from the impending h-file invasion. <So the e7-square is not only self-blocked by the Black rook residing on this square, but the White f6-bishop also covers this important square.> If Black captures the bisop, then the self-blocking rook results in mate. If Black moves the rook, then the f6-bishop covers e7 and the result is again mate. <This double coverage of e7 is what Black (Morovic) missed.>