Axe Fx III and FM 3

  • I had the Kemper Stage and really had a couple of reliability issues. So much to german quality. So it returned the Stage. Now I have the FM3 and are really happy.


    I like the sounds and especially the play feel on the FM3 much better. Kemper sounds good for Highgain for my taste, but i like the FM3 much more for the Fender, Dumble and other Overdrive and Clean sounds.


    Regarding usability the fm3 is a totally different animal than the kemper. With the FM3 you get an editor which works perfectly. Softwaredeise the Kemper is years behind the others - and didnt' make much progress in the last years.

  • ... and it is my understanding that they discontinue support for older models as they go along meaning those models will quit getting any new features. Another issue with their modeler architecture is that software updates with new features frequently change the tone of the patches.


    I don't deny they make a great box with great features; however, when it comes to ease of use (live), ease of getting a good tone, price, and longevity, the KPA really takes first in my experience.


    If you are a tweaker and your sound requires lots of different efx arranged in interesting combinations, the Axe III is pretty much light years ahead in the industry.


    I always think of the KPA as a replacement for a tube amp and stomps. Since I spent decades perfecting this work-flow, it is still the one I am most comfortable with (and I am an EE that can write software).

  • it is my understanding that they discontinue support for older models as they go along meaning those models will quit getting any new features.

    That's standard practice for a lot of companies, though. Discontinued support means the company stops supporting a product, which applies to new features. It's especially prevalent in the software industry(eg. Windows and MacOS).

    Another issue with their modeler architecture is that software updates with new features frequently change the tone of the patches.

    New features don't necessarily change the tone. Changing the modeling algorithms can.

  • That's standard practice for a lot of companies, though. Discontinued support means the company stops supporting a product, which applies to new features. It's especially prevalent in the software industry(eg. Windows and MacOS).

    True. I guess that having product cycles that were 1-3 years makes this a little more of an issue for prospective buyers. I suspect that people would have a little more grief if MS quit supporting Windows 7, two years after they introduced it, people would have had problems with that too.



    New features don't necessarily change the tone. Changing the modeling algorithms can.

    I have friends that complained about it back in the Axe FX II days.... but I wasn't using a Kemper then.... I was using a VHT UL. I can only assume that they were truthful about their experiences.

  • True. I guess that having product cycles that were 1-3 years makes this a little more of an issue for prospective buyers.

    The life-cycle of the Axe-Fx II was 7 years. It was released in 2011 and discontinued in 2018.

    I have friends that complained about it back in the Axe FX II days.... but I wasn't using a Kemper then.... I was using a VHT UL. I can only assume that they were truthful about their experiences.

    I'm not implying your friends are lying. If they're claiming that new features change the tone of Axe-Fx presets, they're simply mistaken. For example, the introduction of the Tone Match feature didn't change the tone of existing presets, nor did the introduction of the IR alignment tool in the Cab block or the Solo band feature in the PEQ block. Changes to the Amp block (eg. modeling algorithms) can and sometimes do affect the tone.

  • It's my impression that few people have had both axe FX units and Kemper and properly tested them before forming opinions.


    Personally, I don't treat fractal units much differently to real amps. It's pretty easy to get great tones, that way, at least for my taste. With Kemper I typically profile the analog counterparts of similar signal chains. The cool thing about fractal is I can tweak fine details that typically require a lot of work in the analog world.


    Also, I can understand possible frustration with the core amp sound of fractal units changing. But the upside is changes that are often improvements to most. Not that Kemper hasn't had its own improvements. They just typically don't involve amp sounds (even though you could argue aliasing improvements did improve core amp sounds, I guess).


    Additionally, there's certainly big numbers of people who don't mind/care for possible profiling improvements (which maybe wouldn't even have to affect old sounds). But if I can EQ match Kemper profiles to source and get measurably closer -- feel this in my hands, too, playing the guitar, prefering the result -- I'll surely be happy with improvements where technically possible. I just don't expect them, neither think sending tests would contribute to much.


    On that end, one reason I'm on the fm3 wait list is fractal's push for improvements in the amp modelling section. There's always going to be some EQ deviations from amp to amp, fractal modeling their own amps, not yours, but I find this kind of development-constant push interesting, somewhat like in the analog world. Certainly there's no moral obligation to feel similarly.


    If it comes to fm3 vs axe FX 3.. if money wasn't an issue, my decision would be based on how much I needed EQ matching, personally. The eq matching of axe fx 3 can narrow the gap between built in models and other examples of these amps nicely, provided the differences are mostly EQ based.


    And yes, updates don't often radically change the amp modelling either... But stacking up developments can add up. Amp modelling improvements have made a very real (and positive) difference in my use of axe units. I'll also be interested to see what happens with neural dsp Cortex too, as I like the workflow of profiling amps as well.

    The bonanza

    Edited 3 times, last by Dimi84 ().

  • I am absolutely in that camp that I've never used a Fractal product hence I'll never diss them :)


    I totally agree with what you've said except about KPA users not wanting improvements. Would I want the KPA to sound better? Of course I would but there are a few other considerations:


    1) As everyone knows, the KPA is about profiling. Once you reach 100% accuracy it technically can;t sound any better. In Modelling ( min my opinion) its not about accuracy but the best sound, so theoretically its possible to be better than the reference amp. I know the KPA isn't 100% accurate but its close to being indistinguishable in blind tests, so I think the law of diminishing returns applies. Therefore we need to be clear on which bits actually need improving. For me that isn't features ( like Dual amps) or more tweaking options but to the base level profiling. The only area that still gets comments is the Valve amp "thump" but this seems difficult to pinpoint exactly.


    2) As per point 1, Kemper have stated a few times that there isn't much left they can do to improve the profiling process so we may have hit a natural limit.


    3) Would I want to pay for a new unit for 1-2% gain if that was attainable? Maybe, but it would have to be tangible. Ultimately I think this is the difference, Fractal have found improvements but at a cost of processing power etc. Kemper have not. One approach is not necessarily better, just different.


    So its not a question of not wanting improvements, its are they possible, what would they be and how much would it cost ( new unit, diverted resources away from OD's and other developments etc.).


    What a great time we live in to face these choices!

  • I totally agree with what you've said except about KPA users not wanting improvements. Would I want the KPA to sound better?

    I mentioned people not minding/caring about possible improvements to profiling itself. So I don't necessarily disagree that a law of diminishing returns applies, for many. It surely does, yea.


    But in so far as technical limitations, I don't know if what CK meant at the trade show is that he recognizes some area for improvement, but that there's such limits (not considering profiling multiple stages)... or whether he seems some minor possibilities, without this being a road worth treading on, like some improvement on kempers' EQ matching tech.


    We could surely go into more detail about profiling. But that's been done a gazzillion times in the past. There's quite specific areas people have pointed to, with tests and considerable detail.


    That aside, on the modelling topic, I think there's many areas where fractal models have sought more accuracy. Modelling is meant to accurately replicate mostly the analog world, right? That's a big part of their focus. I've had amps where the respective amp sims get closer in terms of tweaking in digital vs analog world. It's some times hard to test, considering fractal uses their own amps, but I've seen notable cases of this nonetheless.


    A problem for me -- when it comes to replicating a source tone -- has been that EQ is a weird animal, amp tolerances are a real thing, potentiometers are voodoometers, oftentimes in their variance... And ax8, FM3 have no automated way to bridge that gap akin to profiling. (Well, shooting IRs can go a long way, but EQ matching has several advantages, imo, and profiling does considerably more too).


    Money not being an issue, as well as portability, I'd myself go for axe FX 3 over FM 3 largely due to the EQ matching option. It's just that FM 3 modelling (haven't used an FM3, but axe FX 2-3, ax8 yes) is probably good enough to where shooting IRs is usually sufficient for me.


    Still hard to let go of Kemper though... so I don't think that's going to happen, even if I've gone through quite a gear unloading phase, from gazzillion guitars to Amps, mostly keeping gear that's proven essential to me.

    The bonanza

    Edited 2 times, last by Dimi84 ().

  • In Modelling ( min my opinion) its not about accuracy but the best sound, so theoretically its possible to be better than the reference amp.

    A model is a representation or simulation of something. The accuracy/authenticity thereof determines how closely the model conforms to the original. Good meteorological models, for example, accurately simulate the weather. As such, accuracy/authenticity is important. Of course, not all amp models are based on actual amp designs, but for those that are, what's the point of creating a model that's based on a real-world amplifier if the goal isn't to be as accurate/authentic to the original as possible?


    It should also be mentioned that the Kemper uses amp models, as well.

  • A model is a representation or simulation of something. The accuracy/authenticity thereof determines how closely the model conforms to the original. Good meteorological models, for example, accurately simulate the weather. As such, accuracy/authenticity is important. Of course, not all amp models are based on actual amp designs, but for those that are, what's the point of creating a model that's based on a real-world amplifier if the goal isn't to be as accurate/authentic to the original as possible?


    It should also be mentioned that the Kemper uses amp models, as well.

    To be precise, kemper does not uses models, to the ones that know maths, kemper is reproducing the transfer funtion of a mic in front of an amp, using the mic as an observation feedback for the output, with a known signal over time. It is incredibly genious, but still difficult to achieve due to the nitty gritty of Z domain for digital transfer funtions.


    Modelling an amp is more in line with actually using mathematical models, where they simulate the electric chain with basic mathematical models, and start tweaking the model element by element until the same input, generates the desired equal output under the same conditions, even after modifications on the variables of the math model.


    Two worlds completely appart, Kemper give us a blackbox that reproduces as intended, without knowing whats inside, while fractal/bias/helix, go on modeling even the internal components, where technically you can modify the capacitors and tubes on the supposedly modelled amp, but at the cost of negligible less accuracy, as this is more a trial an error aproach.

    The answer is 42

  • Yes, it does. This is a quote from Christoph Kemper:


    "Profiling is a sort of automated modeling, where a very complex base model of a generic guitar amp is filled with settings that finally form a specific amp model." -Source

    I feel like this can be seen when using the crunch profile in a blank performance slot. Figuratively speaking and partially literally.

  • Yes, it does. This is a quote from Christoph Kemper:


    "Profiling is a sort of automated modeling, where a very complex base model of a generic guitar amp is filled with settings that finally form a specific amp model." -Source

    Regardless of what Christoph was attempting to relay with his words, Alfahdg is completely correct (and obviously a fellow electrical engineer ;). Profiling is NOT modeling at all. The two approaches are as different as a microwave and a stove. Both heat things up, but don't share anything at all in common with regards to the way they do it.


    Additionally, you are quite incorrect about the time between Fractal models ( I may have even missed a couple):


    Standard: 2006.

    Ultra: 2008.

    II: 2011.

    II Mark II: 2012


    II XL: 2014


    II XL+: 2015


    III FX: 2018

    I suppose if you are not a gigging musician, it doesn't matter that much if your entire setup of tone changes from firmware version to firmware version. For me, I would be absolutely furious if my tone changed because of an update. I have spent endless hours and gigs honing exceptional tones that work within the context of a live band. That would be totally unacceptable.


    Now .... what about all the Kemper releases:


    Kemper KPA: 2012


    Hmm. That's it. Of course they released a powered version (no tone changes), a rack version (no tone changes) and a foot controller version (no tone changes). They also released a foot controller (again no tone change).


    I would argue that the Kabinet actually DOES change the tone, and even this completely understandable tone change has likely caused those who were setup for either a FRFR or another cab likely had to do quite a bit of work to get their tones where they wanted them with the Kabinet.


    Wanna make any bets on when we will see another non-firmware compatible Axe release?

  • Regardless of what Christoph was attempting to relay with his words, Alfahdg is completely correct (and obviously a fellow electrical engineer ;). Profiling is NOT modeling at all.

    I never said profiling is modeling. I said the Kemper uses amp models. The profiling process gathers data about an amp. That data is then used to automatically populate a complex base model of a generic amp with settings that form a specific amp model.

    The Axe-Fx Ultra was just a souped-up version of the Axe-Fx Standard. They both use the same amp modeling, thus there's no audible difference between the two. As far as the Axe-Fx II is concerned, all of the models (ie. II, Mark II, XL, XL+) are compatible with each other, and there's absolutely no difference in the sound quality between them because they all use the same modeling. A preset created with the original Axe-Fx II will sound identical on the Mark II, XL, XL+ and vice versa.

  • A model is a representation or simulation of something. The accuracy/authenticity thereof determines how closely the model conforms to the original. Good meteorological models, for example, accurately simulate the weather. As such, accuracy/authenticity is important. Of course, not all amp models are based on actual amp designs, but for those that are, what's the point of creating a model that's based on a real-world amplifier if the goal isn't to be as accurate/authentic to the original as possible?

    My point is with Modelling it doesn't need to stop at the reference amp, profiling can only ever mimic what its presented. Yes I know true modelling is to create an accurate copy. I didn't say accuracy wasn't important but you can make improvements e.g. more gain available than the real amp or less noisey etc.


    I also added " in my opinion"...

  • My point is with Modelling it doesn't need to stop at the reference amp, profiling can only ever mimic what its presented. Yes I know true modelling is to create an accurate copy. I didn't say accuracy wasn't important but you can make improvements e.g. more gain available than the real amp or less noisey etc.


    I also added " in my opinion"...

    With Kemper also, gain can be increased above the amp's regular range (not that you don't know this -- and cool things can be accomplished with KPA that way) at least after profiling. Or with axe FX you can modify the amp sim in such a manner to get more gain than the stock model left untouched, certainly.

    The bonanza

  • Here's one thing though that could be another consideration with axe FX 3 vs fm3... Cliff has said he's waiting for a patent on some "super wow" tech. What that is, whether it'd be useful to me... I know not. But I also don't know if that would require new hardware, be available in both FM3 and axe FX 3, or not...

    The bonanza

  • My point is with Modelling it doesn't need to stop at the reference amp, profiling can only ever mimic what its presented. Yes I know true modelling is to create an accurate copy. I didn't say accuracy wasn't important but you can make improvements e.g. more gain available than the real amp or less noisey etc.


    I also added " in my opinion"...

    You didn't say accuracy wasn't important. You said modeling isn't about accuracy. If we're talking about creating a model of a specific design, it is. If we're talking about modifying an existing design, then no.

  • Whoever cracks being able to profile chorus and modulation effects will be my hero. ?

    Per latest neural DSP video I saw, cortex will be able to “capture” specific pedals themselves, which can then be used as blocks. But I don’t think this pertains to such effect pedals.

    The bonanza