There's an animation principle called Staging (with equivalents in film & theater). The main idea behind this principle is to simplify & amplify to convey the desired "point" of a scene more cleanly and vividly. I think this principle applies to gameplay mechanics in a very interesting way that I don't see discussed very often.
A big part of staging, especially when applied to things happening over time, is what I call Discreteness. Animation is heavily focused around creating discreteness in all sorts of ways - animators will typically remove in-between states and positions to exaggerate motion and break it up into more digestible, discrete "points" (usually key frames themselves) within the movement.
Think of a dash in anime - the characters will often outright teleport from their starting position to their end position to convey that sense of speed. Swinging motions are a good example too, animators will often hold on the maximum point of the swing for a very long time, speed up the point where the object swinging has maximum kinetic energy, and then slingshot the character forward like crazy, all to turn a more continuous motion into something clear & readable for the sake of impact. This even applies to staging - animators will avoid several overlapping actions to avoid taking attention away from what they want to emphasize.
Games are interactive animations at their core, so they will always have staging like it or not. However in games, Discreteness isn't quite as free as it is in animation, because its directly tied to the gameplay mechanics and the core interactive layer of a game.
In games, certain mechanics make it nearly impossible to create much discreteness without undermining what the games are at a fundamental level, while other mechanics have discreteness built into them. I covered this a bit when talking about Burst vs Continuous Movement and Fuzzy vs Discrete Skills in games. To simplify - the more granularity there is in your game's mechanics or general gameplay (like if it's based around subtle management of a lot of different objects), the harder it will be to create discreteness.
I believe there's a pretty strong connection between a game's discreteness & its mass appeal. Discreteness matters because most players have a bad eye for nuanced, subtle interactions - often this requires pre-requisite knowledge of how games work, allowing people to know where to even look. Discreteness in mechanics (again, think of dashing, or state-based attacks) clearly telegraphs that whatever interactions or inputs just happened on screen meant something. This is why high skill gameplay in games with highly continuous mechanics looks easy to onlookers (Quake's LG duels, most racing games) whereas high skill gameplay in games with discrete mechanics looks impressive (precision platformers, combos) even when the precision requirements are the same or lower.
This ties into Derek Yu's idea of Bigness - the more clear & discrete a game's interactions are, the more BIG & COMPLEX people will feel those games are. A game's depth only matters insofar the viewers can perceive it.
Bullet hell games are an interesting case, because dodging bullets is incredibly continuous, but the staging of the games is such that anyone (thinks they) can tell the genre's complexity at a glance. However, once they start playing, they will be discouraged because nothing they do has clear feedback.
This is very relevant to beat 'em ups because the genre is very poorly staged. Their most deep & interesting interactions are subtle and invisible to the average player (crowd control, positioning, decision making), while their most braindead boring aspects (attacks, combos) are the most well staged and clear. As a result this leads to the impression that the games are much more shallow than they actually are.
It's also why I believe things like combos, parries, dodge rolls/defensive mechanics, explicit weaknesses/hard counters and RPG mechanics are so successful - they create very strong discrete layers that make the depth of games more digestible to the average player, even at the cost of real depth/complexity.
Figuring out how to stage the unstageable mechanics is IMO the biggest obstacle most niche genres face. Thankfully staging is multi-faceted. In animation something as simple as a camera angle can turn a poorly staged scene into a good one. In games, smart application of strong kinaesthetic feedback can turn a unweidly gameplay mechanics/dynamics into something much more comprehensible to the average player. Identifying & defining different common behaviors and giving them a specific term helps a lot - like point banking & restreaming in shmups. Gotta get creative here.