FractalAudio a écrit :
This is a common misconception. The Kemper does not use "3-dimensional" IRs (whatever that is). The file size if far too small to support that. It uses standard IRs and quite short ones at that. IIRC they are only 256 samples. The Kemper has seven (last I checked) base amp models. It finds the model that is closest to the measured response. It then does an EQ and gain match for that model. It also finds the input EQ. The heart of the Kemper is gain and EQ matching. You'd be surprised how far that goes in recreating an amp.
FractalAudio a écrit :
Last I checked it was seven base models. There may be more now. Seven would be more than enough IMO. Most amps derive from a handful of base designs: Fender, Vox, Marshall.
The Axe-Fx II does not rely on base models. Each model is unique.
Gain matching is fairly simple but I don't want to disclose how it is done as it is part of a patent I have pending.
The main point was the misconception that the kemper uses dynamic convolution or some kind of different IR technology. It does not (the file size is much too small to support that). In fact you can open a .kipr file and see the IR data in there. It's encrypted but it's clearly there. IIRC there are two 256 sample IRs. My guess is that one is the input EQ and the other is the output EQ. As these short IRs lack sufficient low-frequency resolution I would also guess that low-frequency resolution enhancement is being done using IIR filters although other techniques are possible. IR technology is great though. Anyone who thinks otherwise just doesn't understand how powerful it is.
