We can identify the outliers while at the same time observe the strong positive correlation between x and y.
Even though they look similarly, I believe the logical reasoning behind why scattered plot is superior in this case is because it focuses on showing the locations of each individual point, rather than the binned counts, thus it's more suitable for the purpose of identifying outliers.