I checked on the kernels situation. I did some preliminary tests with increased kernels when sampling, and in the case of the tape, 2 kernels actually sounds "better".
Why is that?
At 2 kernels, Saturation is more pronounced / effective. I did some tests on 10 and 7 kernels. First, there's some major clipping artifacts when pushed, and also the saturation effect is lost, and the quality not really that much different - maybe lows a *little* tighter. So maybe with something like an old tape machine, the desired effect is more pronounced with lower kernel usage. I'm still learning quite a bit about how NAT works, but from my initial tests, this is my experience.
I'll continue to play around with it and let you know if I make any other observations.