To me the natural progression or next step at the flick of a switch gives me this...
A fully customizable Avatar and portrait where the player can effectively create any 'face' they can imagine, including their own.
I think the baseline for achieving this is somewhere around 200 archetypal heads (100 for each gender) and say a dozen body phenotypes, then modular variation from there. So going up here by at least an order of magnitude.
Sort of an in-game version of Maya light, that can be used by anyone without needing expertise in design. A tool created by artists for use by non artists, which nevertheless makes the player feel like they are the creative force behind designing their characters look.
This sensibility would also be brought to the outfits, for the dress em up aspect
All this visual stuff would by heavily stylized so that game does not appear dated in 5 years time. Here I'd look to 2d animation more than 3d as the goal for a vibe, and trying to achieve something timeless like that, not the cutting edge of mimesis per se. Since the goalpost there is always moving, and making it look real isn't as important as making it look 'fantastic' if that makes sense hehe.
Part of what makes BG3 cool I think is that it is already doing this. So just pushing that as far as it can possibly go for D&D.
Things like voice, where having options for different cadences, including voice modulation would do the same for sound. Also a gesture and emmotive suite, which could make two characters who are otherwise identical feel completely different just from that stuff alone.
For me environmental exploration would start very basic. Goal here would be to capture all the mundane stuff like climbing, swimming, tumbling around and make that feel so cool with interactivity that it throws you right back to your elementary school days on the playground or building tree forts. There should be no tree we cannot climb hehe.
That way when you actually get to the truly fantasy stuff, and abilities beyond, it feels totally next level, since the basics of exploration adventuring would already be blowing your mind anyway at the floor. Making even the lowliest cantrip feel high magic by comparison, because the basics already felt like magic, before you even hit your first level as a proper adventurer. I just picture whatever the best in class is for a modern FPS there, with that sort environmental interactivity during exploration, but then extended to the entire party, all using those features - as a group. Party on!
Ps. For inspiration, when the ropes start looking like this in Baldur's Gate, we'll know we're getting there... lol