Depends
But, the biggest difference to my mind is levels. More specifically, dynamic range. While some people might be listening in a high end HT environment, the same mix has to sustain a lower listening level for a more casual listener. I'll mention now though, that some HE mixes are VERY similar to the cinema mix. If the delivery spec allows, and the original mix is quite controlled, it might play great pretty much as-is.
But let's assume we're starting with a cinema mix that's just too wild........
The most cardinal sin a mixer can commit is sending out a mix that's too dynamic and people have to reach for their volume controls to turn the loud scenes down / quiet scenes up. What
they'll say, probably, is "I couldn't hear the dialog, the mixers don't know what they're doing and are making the dialog unintelligible. It's unacceptable." Of course the quality of the original recording and performance is probably the most important factor here, but having mushy dialog AND wide dynamic is the likely to cause negative feedback from the public.
So, dialog
quality aside, part of that story is there will be an upper limit to how dynamic a mixer feels they can "get away with" without upsetting too many people. Although, each mixer will have slightly different personal preference for where they feel that limit is, and moreover, some films justifiably warrant a wider dynamic than others.
Moving on from mixer preference, the distributor's requirements may come in to play in determining the dynamic range. For example (for HE) Netflix request a certain measured dialog level and certain peak level. Although in Netflix's case the range is quite generous compared to others, and there's not too many criteria set, this will still dictate the maximum
permissible range between dialog average and the loudest sfx/music.
I mention Netflix despite their relatively relaxed spec, as they make it more publicly available than most. Section 4.2.1 in this document describes those limits:
https://drive.google.com/a/netflix.com/file/d/0B37xotBvlV3Qc1E4VW5sMzNwcmc/view?usp=sharing
But like I say, it's not uncommon to have tighter requirements than this, precluding the use of wide dynamics. Quite often the same mix goes on Blu-ray and VOD, even if the final encoding is higher bitrate on disc. I guess they base the spec on their target audience. Not for Atmos, but I know some US broadcasters are still requesting things to be done with no wider dynamic than old analog TV broadcasts, so while not totally relevant, it shows they feel this is appropriate for their listeners, and I don't necessarily disagree, even if it does feel very flat in the studio or a HT room.
Anyway, aside from getting the dynamics down, there's quite a lot of perhaps less noticeable (but cumulatively worthwhile) tweaks that go in to making an HE mix really work. A few things that might get tweaked as against the cinema mix are:
- EQ - brightness/boominess might be different on dialog
- Bass management - As per this thread it's brought in to play for the first time when the mix goes to the HE stage, do we need to redirect bass at all in the mix? Maybe.
- Ambiences might need lifting in quieter scenes
- Music might need lifting
- Surrounds from the bed play as point source in HE, so might need taming/diffusing if they're distracting
It's probably a very long list in reality, but that's the sort of thing... I probably do less HE than
@Soundmixer though, who may have some more thoughts.