So it turns out anti-gravity gain was the cause of my tuning problems.
Glad you got it sorted out.
It seems to be a fully functioning system that uses gyro input to stabilize, not just higher min throttle.
Of course it's a fully functioning system that still uses the gyro, it's just that the PIDs get
"thrown out the window" because you've exceeded their ability to properly function at too low of a level.
Some command inputs become unstable (which is to say basically either ineffective or producing highly undesirable results) at lower throttle levels without airmode enabled because there's simply not enough available reactivity at lower throttle levels to execute those commands. And, yes, all it basically does is to
momentarily raise min throttle just enough to perform the command function adequately.
It works like this: Let's suppose you're at 10% throttle. You now input a command that requires a 30% reaction from the props. The reaction is a function of affecting one of props set by 15% positively while
the other set gets 15% negatively. At only 10% throttle you do not have the command authority needed to
impose a 15% reduction of spin on any pair of props, so air mode bumps up the min throttle by 5%
automatically so that the opposing prop sets can adequately impose the required 30% differential required
to satisfy the command input.