Let us not involve event horizons here. They muddy the waters. Under general relativity, there are situations where gravitational time dilation can be considered as "real" time dilation.Does this mean that in GR we can talk about the "real time dilation independent of observations"; contrary to SR where the real time dilation has no physical sense and we can only reason in terms of what we can actually observe?
Under special relativity (and inertial frames), time dilation is always due to relative velocity. Both observers consider their peer's clock to have slowed down.
Under general relativity, there can be time dilation due to gravitational potential. An observer higher looking down at his peer will consider the peer's clock to be running slow. The lower observer looking upward will consider his peer's clock to be running fast. If you wait long enough, an unambiguous difference builds up between the respective clock readings. This could be thought of as real time dilation. It is physically measurable and does not arise due to differing choices of simultaneity conventions.