new "special kind of profiles" (just dreaming....)

(sorry if it’s long, or already proposed long time ago)

I know this is normally the job and philosophy of a simulator and not of a profiler, but I’d love to have an additional mode, with all the knobs of the real amp, and their real behavior in the kemper.

Imagine an amp is profiled in all of the possible combinations of all its control knobs positions (not really all the positions but a discrete sampling of them). Then, the profiler (or a computer via rig manager if it's to hard for some reasons with the profiler itself) would merge this huge amount of data into a smaller file associated with a code that can recalculate a single profile from any knob combination chosen by the user. (It could even interpolate the results for knobs values that fall in between the ones originally profiled). It would be done.

In my mind it’s analogue to interpolation of a curve using a 1D function in mathematics, but with the 2 differences that it is for all of the parameters of the profile (which I see like a set of constant parameters fixed during the profiling procedure) and that it’s a multivariate fitting (one variable per knob). In other words, It can be seen like morphing but not changing the front panel accessible parameters like in the normal morphing, but changing the fixed parameters hidden in the profile itself.

We could then exchange or purchase such a “special kind of profile”.

sorry for asking this again !

If I can bore you a little bit more...

I know well the problem of the interdependence of the control knobs, and my idea was precisely to overcome this problem.

Let me explain :

Rigorously, you would have to perform one normal profile for each possible combination of the knobs of the amp, which will quickly represent a huge amount of profiles : if you have 3 knobs and you want 10 samples between min and max, it makes 10^3=1000 profiles to perform. But if you take only 5 samples between min and max for each knob, it makes already less, 5^3=125 profiles. In the general case, you have (profiles to perform)=(samples per knob)^(number of knobs).

Starting from these profiles series that span all the possible settings of the amp, what I imagine that it is possible calculate (or interpolate) the profiles for the knob positions/combinations you didn’t physically profile (but between the ones you profiled). It can take the form of a much smaller file used to recalculate the real profiles (you don't use the thousands of profiles anymore). The principle I talk about is basically the same than what is done when dealing with experimental data : you have a set of experimental data (herein : profiles), that you fit to a mathematical law (the thing to find here), and you end up with only a few parameters that allow you to interpolate between the experimental points. With only two numbers only you can recalculate all the points of any parabola for instance, here it's the same, but multidimensional, and instead of recalculating parabolas, we recalculate profiles.

This was technically speaking.

Now imagine it works. For a 5 knobs amp (gain vol trebs mids bass) and 5 samples/knob, it’s 5^5 = 3125 profiles to perform. At say 5 minutes per profile it makes 260 hours of work. I know it’s a lot, but it’s done once and it can be sold afterwards. Don’t know if it’s economically realistic. Personally, if it works I would pay 30 to 50 for one well done amp.

I'm sure guys at Kemper must laugh when they read what seems so easy to customers that have no idea of what is really behind a profile. But who knows... To what I imagine they are, profiles are just sets of constants used in the machine to digitally normalize, convolve, filter, etc... the incoming signal. If that's true my idea should be feasible.