Les Paul mods: 47nF vs 22nF vs 15nF tone capacitor

Hi folks. If you own a guitar, you are probably aware that there are a bunch of customizations that you can do with the passive circuitry of your favorite axe.
There are lots of websites that shows and explain different mods, but it's hard to find one that shows some data or nice graphs. So, inspired by this forum thread, I did simulations for some Les Paul mods (I own an Epiphone Les Paul Custom) and I'll post some results.
This one shows the effect of changing the value of the tone capacitor. Popular choices are 47nF, 22nF (these are the ones I found in my guitar) and 15nF.
They usually say that smaller caps give brighter tone, but how exactly do they affect you sound? Here I'll show you.
The graph is a logarithmic plot of the frequency response from 20Hz to 20kHz between the pickup ideal signal and the output, for the three values of the capacitor, and for different positions of the tone potentiometer (I set ten values of a logarithmic sweep from 1Ohm to 500kOhm).

As you can see, the different capacitor values are non-influential when the tone knob if fully on the bright setting ("10" position), but the more you close the tone, the more the effect of the capacitor becomes apparent: with lower cap values, the resonant peak for the dark tone position shifts higher in frequency and becomes more pronounced.
This holds true at whatever volume position: these plots are for full volume setting, but I could have chosen another volume position as well.
So the bottom line is: smaller caps give brighter sound, but only when the tone knob is on a dark setting. In other words, a smaller cap limits how dark the tone can be. This can translate in being able to use the full range of the tone control without the sound getting "muddy".
Looking at some forums, it seems that many people like a 15nF for the neck PU and a 22nF for the bridge one.

Of course, this is a circuit simulation whit ideal components; if you want to know how this translates in reality just open your axe, check some different caps, and - as always - let your ears be the judge!

PS:
The software I used for the simulation is Qucs on Linux, but on the website you can find downloads for Windows and Mac OS X as well.
There's no potentiometer component in Qucs, so I just used two resistors with the values bond together by the equation Rvb=500k-Rva.
For the pickup I chose values that, combining various sources, seem to be good representatives of a standard Gibson PAF pickup.