Don't get too comfortable with this firmware. I had an epiphany a few days ago and have been crunching equations and coding like mad since then. I realized I wasn't adjusting the speaker impedance curve to coincide with the compression. As the speaker compresses its impedance curve changes. Makes a big difference. Tighter low end and clearer highs. The math is a bit of a bear but I finally cracked it this morning.
I'm really excited to see the development moving towards speaker/cab simulation. I always felt it's (or was) the weak link in modeling. Can't wait to try it!
I wonder if this impedance curve variation depends only on the compression amount or if it changes with different speakers' attributes (eg: speaker diameter, voice coil diameter, coil wire awg, magnet composition, etc.)
I Imagine the definitive speaker/cab sim as a tab where the user inserts all the basic characteristics of a speaker, number of speakers and cab dimensions and the algorhythm calculates all the non-linear behaviours of that particular configuration based on the output power of the selected amp.