Full profiling

  • Hello to the Kemper team,


    Do you think it will ever be possible to fully profile an amplifier?

    Let me explain: If we consider that one profile is the equivalent of a picture of an amp on a particular setting, could we imagine to make a video of this amp with all the possible settings?

    The reason I ask is that I have lost countless hours in trying thousands of profile to NEVER find what I was looking for. And yes I tried to tweak, add EQs, changing cabs etc... and it doesn't react at all like the amplifier would have done if I had turned the knobs on it.

    It is SO frustrating to always have to deal with settings done by someone else.


    Do you think you could invent a protocole that would allow the KPA to make complete profiles? Perhaps something where the user would have to progressively turn all the knobs of the amp so the KPA would analyse how the amp reacts? That would be so great...


    Thanks for reading me!

    Edited 3 times, last by Xisale ().

  • Have you ever profiled an amp at a setting that you personally like?

    If not, please do it once because this is where the solution to your problem is.

    From there you'll realize that the vast majority of amps don’t have infinite sweet spots so that the ones they have are easily profilable as well.

    If this is true for the most complex amp of the ones I own (a Marshall JVM Joe Satriani) this is the more valid for any amp.

  • Ingolf: If only I could!! I would love to do that but I don't own the amps I'd like to profile and I don't know anybody who owns them.

    I even tried to look in the shops and backline rental companies in Paris (where I live) to see if there was some and they are f*****g impossible to find. So my only options are to download profiles from the rig exchange or to buy them. I tried both, and although some of them were quite close to my taste, I never found exactly what I was looking for.


    If the people who made the profiles could profile the "whole" amp and not only one setting, this would definitely solve the problem.

  • Hi Sambrox!


    I actually tried that. Tim from Livereadysound was nice enough to make me one (Hi Deadpan if you read this!)

    It was actually really good but still not exactly right on spot though (pretty hard to hit the center of the target at the first shot I guess). We are thinking of solutions, but I don't want to annoy people too much with that.


    Michael from BHP also made me a great one too for solos some time ago, but I think he is pretty busy nowadays.

    Edited once, last by Xisale ().

  • :)

    I keep thinking about my "video over picture" idea, though.

    Imagine that you'd make a "video profile" of an amp.

    For example, you would have all the EQ knobs in the middle position, and while the KPA is making the profile, you would slowly roll the Mids knob in a way or the other. So in the end you would get a "all the mids setting possible" video profile (which means a profile that has a length in time) in which you could pause wherever you want when you like the sound. It seems pretty realistic to me, especially considering that the current profiles use almost nothing for storage.


    Having the whole amp settings possible in one file would obviously be even better, but if you'd tweak many things in a long "video profile", that could work imo.

  • Well after all profiles are snapshots and amps are drasically different with different master volume setting and so on.


    This is why i make D.I. profiles with all relevant gain and mastervolume setting. Still the EQ is a factor that i do not cover that much that comes down to my taste and cab but i include often a "noon" version ...


    So give it a try if you like to, they should cover the whole amp well. Look for Bommel in Rig Exchange.


    But i know your frustration that is why i had to made so many profiles. capturing every knob and so would be ultra cool. IT would be enough if ther are more eqs gain knobs and mastervolume setting inside the kemper modeled after the most important amps ... that way everything could stay the same but especially the poweramp sim would save a lot of time

  • Marshall2550: I tried that too, but I didn't find any studio which owns a BE100 or a SLO100. These are pretty rare amps.


    xzyryabx: OK my bad then. Did you have the same idea of a profile with a time length in which you could pause? I have searched your name in the research function and didn't found the topic.


    Bommel : I have actually already tried most of your profiles. They are among the bests imo! I may actually chose one of them for my studio tone. So thank you so much for giving all of these profiles for free! It's greatly appreciated.

    Btw: I tend to think more and more that "everything at noon" is the best setting for an amp :) Maybe slight adjustments here and there but if you dont like the sound of the amp with everything at noon, then the amp is not for you.

  • For example, you would have all the EQ knobs in the middle position, and while the KPA is making the profile, you would slowly roll the Mids knob in a way or the other. So in the end you would get a "all the mids setting possible" video profile (which means a profile that has a length in time) in which you could pause wherever you want when you like the sound. It seems pretty realistic to me, especially considering that the current profiles use almost nothing for storage.

    unfortunately that’s not how Rome knobs work though. They are all interactive so the position on the treble knob also affects the bass and mic knobs. Also the gain and master volume positions affect the tone as well as affecting each other. Plus all of those have less effect than the speaker and mic type and placement so the idea is nice in theory but totally impractical IMHO

  • unfortunately that’s not how Rome knobs work though. They are all interactive so the position on the treble knob also affects the bass and mic knobs. Also the gain and master volume positions affect the tone as well as affecting each other. Plus all of those have less effect than the speaker and mic type and placement so the idea is nice in theory but totally impractical IMHO

    I am conscious that all the knobs are interactive. But this doesn't contradict my idea.

    If you'd make this "all the mids possible" profile, then you would have all the reactions of the amp with all the mids settings possible (and the rest at noon. Or not) in one profile.


    Imagine you are watching a video of a guy trying an amp on Youtube, and while he is playing, his friend is rolling the knobs on the amp, and at some instant you think "I like what I hear there ! Too bad he didn't stop on this setting".

    With what I'm suggesting, you could do it.

    That's all I am saying.

  • I understand what you are asking but the interaction is real time so simply having all the positions of the Treble knob while the B and M are at noon isn’t enough. You need to have all the possible positions of all knobs with all possible positions of all the others.


    The solution to what you are asking is traditional modelling like Fractal use. the results if done well should be as (or more) convincing than a “video” profile. The downside is the amount of computer processing required. The beauty of the Kemper approach is how little processor power is require by comparison. I can’t envisage anyway to achieve what you ask without massive increases to the KPA processing which would presumably require new hardware and even then you still have the issue of trying to capture all the knob position possibilities. At the end of the day I would just by an AxeFx for that level of user programability

  • In my personal opinion this would only make sense for direct profiles without the additional interaction from the speaker or microphone. From a technical point of view this would require a whole bunch of measurement points. I'm sure, that you don't need every .1 step in between the EQ settings but you would need enough measurements to capture the behavior and interaction of the knobs.

    The challenge would be do develop a kind of learning algorithm which learns how many and which measurements are really needed to capture the amp. If you only capture every second EQ value and one capture would take about 45 seconds, you would need about 2 hours to do this, without the time to document, setup and so on. This time could be reduced if the algorithm is smart enough to guide you through the capture process.

    But there is another problem, not every amp is equal, some have presence, some have resonance and so on, this could lead to many combinations and situations which you have to go through.

    I think the problem is not that it's not doable, the problem is more if you can develop a way to bake this into an interface the user can use. What I can image first is a way to capture the gain by multiple measurements.

  • I understand that. I really have no idea of what the KPA is able or not able to do of in terms of processing.

    In my opinion, the Kemper sounds more "realistic" than the Axe FX, plus I really like the fact that it just respects the mics you are using and how you place them in front of the cab, etc...

    But to me, the augmented reality of the KPA is where it is weak compared to an Axe FX.

    Once the profile is made, I really find that the best option to keep its realistic vibe is to do absolutely nothing.


    I understand what you are asking but the interaction is real time so simply having all the positions of the Treble knob while the B and M are at noon isn’t enough. You need to have all the possible positions of all knobs with all possible positions of all the others.

    This was just an example. you could make a more a less long "video profile". For example, just rolling the Treble Knob could last for 5 seconds, no more. One Kemper profile weighs about 6Ko, which is almost nothing. A video typically uses 24 frames per second. If you'd make a 5 seconds video profile, then you would have a file of 6x24x5= 720Ko. That's not even 1Mo. And you would have plenty of frames (profiles) available in it that you could use.

    You could obviously make more complex video profiles where you would turn the knobs in different combinations and make it last for 30 seconds or so. That would weight only 4,320 Mo. Still not so much IMO. You wouldn't necessary have ALL the different combinations possible, but you would already have so much more choice than what you have when you buy a bundle of profiles.


    If there was a solution to make "full profiles" or "video profiles", then the Kemper would be the absolute perfect machine.

    Maybe the KPA is able to do it or maybe not. Maybe this is where we would need a "KPA 2" ?

  • In my personal opinion this would only make sense for direct profiles without the additional interaction from the speaker or microphone. From a technical point of view this would require a whole bunch of measurement points. I'm sure, that you don't need every .1 step in between the EQ settings but you would need enough measurements to capture the behavior and interaction of the knobs.

    The challenge would be do develop a kind of learning algorithm which learns how many and which measurements are really needed to capture the amp. If you only capture every second EQ value and one capture would take about 45 seconds, you would need about 2 hours to do this, without the time to document, setup and so on. This time could be reduced if the algorithm is smart enough to guide you through the capture process.

    But there is another problem, not every amp is equal, some have presence, some have resonance and so on, this could lead to many combinations and situations which you have to go through.

    I think the problem is not that it's not doable, the problem is more if you can develop a way to bake this into an interface the user can use. What I can image first is a way to capture the gain by multiple measurements.

    Interesting answer! What you describe would be comparable to doing "motion picture". If you need to picture every frame to create the video, then yes it will take ages to make. But maybe there could be a way to profile more like a video camera?

    I have really no idea of what is possible and what isn't from a technical point of view.

    Maybe doing that only for direct profiles would be more realistic indeed. I really have no clue.

  • But to me, the augmented reality of the KPA is where it is weak compared to an Axe FX.

    Once the profile is made, I really find that the best option to keep its realistic vibe is to do absolutely nothing.

    Hm, I see this a little bit different. Especially this point is what I absolutely love about the Kemper. It can bring a sound in complete different direction which was not possible with the real amp. But that's just my opinion because if I want a specific setting of my amp I do so and for me it's more important to get the sound I want than that the it's like the real amp.

    If there was a solution to make "full profiles" or "video profiles", then the Kemper would be the absolute perfect machine.

    Maybe the KPA is able to do it or maybe not. Maybe this is where we would need a "KPA 2" ?

    The idea of a kind of "grouped" profiles is nice but actually the Kemper can take about 1k profiles if I am correct. I you want to have 5 different settings per bass/mid/treble this would need 125 measurements and profiles, and that's only one channel. Let's say you don't need all the meta information stuff, fx and so on in every profile I don't think that you could have more than 10-20 of this on the actual hardware. By the way if one measurement, including dialing in the amp, dialing in the Kemper and so one would take about 1 minute you would need 2 hours just for this. Add 5 more settings for presence and you would end up with 10 hours. ;)
    But I'm with you, if I would ever think about a new design based on the Kemper algorithm I would definitely also think in those directions but I would try to do it smarter than only snapshots.

  • But maybe there could be a way to profile more like a video camera?

    Than you would have to move multiple knobs on the amp at the same time and tell the Kemper at the same time what it actually measures. I would go a complete different way and let the device learn how the amp reacts. Let it learn if there is any interaction between the EQ curves and if how big the interaction is. In some cases you can get away with not so many measurements and calculate the rest by yourself and in some cases you need more information. Speech recognition is a bit similar and got it's quality from training the algorithm.