I have captured the raw output of my amp (using an attenuator and di/reamp box) as well as separately captured the mic'd amp+cab model of the same setup. Bypassing the cabinet in the amp+cab version of the model I captured sounds nothing like the raw amp capture - the frequency response is way off, and not in a way that's easily fixed with some light EQ.
Since the setup for amp+cab capture only uses mics for feedback (no direct output from the amp), it seems the machine learning process that creates the model has to make guesses about what parts of the sound come from the cab vs. the amp, and it gets it very wrong sometimes. For how I use tonex (into a clean poweramp then a regular guitar speaker cabinet), this essentially means only the standalone amp or pedal+amp models have a good chance of being accurate. The large amount of work people have put into making combined amp+mic captures results in inaccurate response for users that want to play through only the amp section of that capture with the cabinet bypassed.
Does IK Multimedia have plans to improve the amp+cab modeling process to include an option to have a direct amp line level input in addition to the cabinet mics, similar to the standalone amp or pedal+amp captures? This would help ensure that both the amp and cabinet models are individually correct when used separately.
