But it is really conceptually rather simple:
delta = change
gain delta = how much gain changes depending on Angle and Distance
freq delta = how much frequency/spectrum/filters change depending on Angle and Distance
time delta = how much delay/phase/time-aspects change depending on Angle and Distance
width = how different L and R channels are in various phase-releated aspects
So if you are doing electronic music and have very bright synths and don't want them to loose High Freqs, as an example taken from a few posts ago, you use lower Freq Delta.
If you want less "time spreading" on transients on things like drums you use closer Distance (as one would normally mix drums anyway), and if you want even tighter, you can lower Time Delta.
The delta values control the relative strenght of various psychoacoustic cues of spatial hearing. The default delta values are appropriate for acoustic instruments IMHO. You can customize them as you see fit. In general, if the goal is to create a unified spatial environment such as scoring stage for orchestral/scoring work, when using multi-instance, you should choose these delta values to be mostly consistent for all tracks. Think of it as setting up your "pan laws", but here it is more "positioning rules" and it is more complex. You can make individual track variations for problem cases if needed. (like you are doing hybrid electronic/ochestral and you have a super-saw synth in your mix and you want it to be ultra-bright, you might want to undo some of the effect of Freq Delta --- it is less natural, but so is a super-saw by it's innate nature --- so as we get into more abstract things such as synths that never really existed in the real-world acoustic space, then basically no one has any idea what "correct" is, bc there is no precident for it, and hence you can define you own behavior as you deem to be creatively radically doodical tm. ) When working with real instruments I suggest to keep deltas similar/equal for all instruments/instances as much as possible and the default values should be a perfectly great choice to use.
Does it help?