Axe FX II and Kemper - some thoughts

[stratzrus]

Originally Posted by greiswig

Great review. I have an Axe II now, but haven't used the Kemper. However, my graduate degree is in Cognitive Psychology specializing in HCI (Human-Computer Interaction). I've been working as a Usability Specialist, UI designer and now User Experience Manager for nearly 20 years. So it is with some authority that I can say...

...if a feature isn't readily usable, it might as well not be there for most people.

True, there are some who are willing to spend hours tweaking things to get just the right sound. Heck, I'm one of them...I've spent so much time tweaking and rebuilding my D-style amps that it's a little scary. So I honestly expected the AxeFX to be easier to do that with. To my surprise, it isn't. But at least I don't worry about high voltages with the AxeFX.

You can get some really good amp sounds out of the AxeFX. Most of the patches it comes with do the unit a disservice, IMHO: they show off the effects more than they show off the amps. But trying to get down to basics (amp->reverb->cab-> out) reveals that even the settings they have on most of the amps don't really show them off for what they purport to be. And what is the difference between the low cut on the transformer versus on the speaker versus all the other places one finds a low cut control? And that's just one example.

I've done enough UI design on enough hardware and software that I know you can accommodate both: have things like default parameters and presets that make it work for 80% of your users, and allow the other 20% to dive in deep and change things. Have a set of presets that are purpose-built to sound like the bare-bones models they represent, then have another section where you show off the effects. It can be done. And I think Fractal would benefit from having someone involved in professional Usability work with them on the UI of the unit itself as well as their editing software.

If the Kemper and other competitors are about as good in the sonic realm and are easier to get the good sounds out of, Fractal may learn the hard way that usability is indeed a key market differentiator. Most guitar players would rather play music than try to figure out what controls in a heavily-laden UI do.

Originally Posted by tubiux

Great reading guys.

@Greiswig: I fully agree with your points.

+1

I think most familiar with my posts would agree that I am fully in the Axe FX camp and think it's the greatest thing since sliced bread but Greiswig makes some valid points. I'm not a designer but the notion that it could be made more user friendly, the editor too, rings true. With that said, Cliff has done a spectacular job and has moved the entire modeling field forward by a quantum leap. In time, as firmware upgrades are offered, I hope that improvements continue to be made to both functionality and user friendliness.

Originally Posted by m~Dan

I studied electrical engineering - but I don't understand most of this stuff... I'm more into embedded systems

But BTT:
What I tried with my Kemper ist taking 2 profiles of an amp with different cabs/micpositions. Those 2 profiles sound completely different when playing with FRFR. Though... if I went into an ENGL 50 watt poweramp and my HUghes Kettner 212 cab and remove the cab in those 2 profiles - the profiles sound VERY close - I think in an A/B I couldn't tell.

What I'm trying to say: You shouldn't discuss that much HOW it could be possible (but it's very interesting though, I love forums for those kind of discussion) but try it by yourself and see that it works actually quite well. Is it a 100% thing? surely not. Is it close enough for me. Yeah, absolutely

Agree.

Dan has said what I've been trying to say but better because it's based on personal experience with the actual unit. The fact that, with the cab removed, the profiles sounded very close proves that the function works very well. The fact that they don't sound the same proves that it's an approximation not an actual removal. Most players would agree that if the two were very close that's all that's needed and as such serves the intended purpose, even if not with 100% accuracy.

Originally Posted by fakeox

Rock(what's Left) is about what feels good not what matheticians can work out with a pencil.

But of course music isn't only about rock and the culture/values associated with it. Rock in it's purest form is about simplicity and modelers, from a design perspective, most certainly are not.

Thanks to all who contributed to the scientific discussion. I found it more educational than any other thread in recent memory.

[paulmapp8306]

Originally Posted by SonicGator

True, but an antenna is nothing more than a "speaker" for RF. Same loading and resistance principles apply and reflective power and frequency intermodulation all play a roll in the dynamics of the RF amplifier to antenna. You get distortion and/or overdrive in an amplifier once you pass the 1dB compression point (Third Order Intercept).

In an RF amplifier circuit, you want to avoid the 1dB compression point because Intermod is potentially a detriment to your communications signal; the equivalent in the Audio amp is overtones, ghost notes, and harmonics, which can be very good and musical.

In Audio amps, when we want to rock, we want to exceed the 1dB compression point, which = overdrive and distortion = rock!

So yes, I believe it is technically feasible to think that an audio engineering firm, in this case Kemper, can analyze a circuit with enough detail to determine what impact a speaker has on the overall tone.

Onl;y got this far in the thread - and held off comenting once but feel I have to now.

there is a difference in you analogy. In the RF world when simulating the effect of antennas and thel ike - you know the antennas properties. With the KPA they dont know which speaker is used so therefore dont know its properties.

IF you could tell the KPA "I used a V30, and an SM57" or "I Used a Greenback and U87" then Kemper could make a decent stab at removing that (nor perfect as mic distance and angle would not be known). the point is they dont.

If you took a head, and played it through 7 different speakers taking profiles of each - then removed the cab part from each profile, for the process to work all 7 profiles should sound the same. While I dont have a KPA (love to get one to try) basic physics says that wont happen.

If you take a forfile from just one speaker you dont really have that comparison to make - and in a single instance the "guess" the KPA makes may well be close enough. It may even sound very good. Im not debating that, but Im completely with Jay on this - its physically impiossible to remove A from B when you dont actually know what A is.

[paulmapp8306]

After catching up on the rest of the thread - all my points have been highlighted - both the impossibility to be acurate, and that the cab vcan be removed from two prifiles taken with different ones - and sound roughly the same.

Carry on

[guitarnet70]

Reading this thread I couldn't avoid to think about what Jay would be able to do with a KPA in his hands...with his knowledge and the technology that he can access, profiles made by him should be something....

Christoph: are you reading this thread?

[MaxTwang]

I was thinking if Jay got his hands on a KPA, a few days later we'd be reading about his profiling experiments to better understand how the cab removal works.

To those talking of too many unknowns to solve the equation, in the years of development Kemper may have done much study on standard guitar speakers and mics and compiled the information to infer which speaker/mic is used in a profile based on certain traits found in the profile that match the data on specific speakers and mics, then remove the assumed speaker & mic. It's not 100% but could have a high probability for success.

edit:
Since only the original profile is matching something that exists, then 'close enough' would be the standard for removing or replacing the cab & mic as we have no point of reference/comparison for the profiled amp w/ new cab. What would be interesting is to profile an amp twice, each time with a different cab - cab A and cab B, but all else the same, then in the first profile use the Kemper to switch Cab A to Cab B and see how close that is to the profile of the amp with cab B.

Anyway, until the Kemper hits our shores, this is the only fun we can have with it. And this theorizing does make me want to play guitar. So I'm off.

[stratzrus]

Originally Posted by MaxTwang

What would be interesting is to profile an amp twice, each time with a different cab - cab A and cab B, but all else the same, then in the first profile use the Kemper to switch Cab A to Cab B and see how close that is to the profile of the amp with cab B.

See M~Dan's post #93 above.

[Jay Mitchell]

Originally Posted by m~Dan

What I tried with my Kemper ist taking 2 profiles of an amp with different cabs/micpositions. Those 2 profiles sound completely different when playing with FRFR. Though... if I went into an ENGL 50 watt poweramp and my HUghes Kettner 212 cab and remove the cab in those 2 profiles - the profiles sound VERY close - I think in an A/B I couldn't tell.

That simply means that the algorithm used to guess which part is the speaker works consistently. It does not establish anything about the accuracy of the guess. You have established that what is left after removing the "speaker" part of two different profiles sounds like an "amp" (the same "amp") in both cases. The relevant questions are:

1. How closely does the "amp" portion match the physical amp when both are played through the same physical speaker?

2. How closely do the "speaker" portions match the sounds of the physical cabs when the cabs are mic'ed the same as they were during profile acquisition?

[m~Dan]

Hm... I don't have an amp here (and no room where I could profile without my neighbors freaking out because they probably think a UFO has just landed).

But I could try it with a preamp and different IRs instead of a real cab ... that should show how good it works too, right?

[Jay Mitchell]

A little tutorial is in order here. It appears that most of the participants here are unaware that there are many response features - thinking in the frequency domain for simplicity - that can be produced by either an amplifier or speaker. These generally fall into the realm of "smooth" response features. They include:

1. Shelving - all frequencies above or below some "break" frequency are present at a different level, either higher or lower, than frequencies on the opposite side of the break frequency.

2. Band boost/cut - a band of frequencies is present at a higher or lower level than frequencies outside of (higher and lower than) that band.

3. Highpass/lowpass - frequencies above or below a given frequency "roll off" (decrease) at a level that is constant with respect to logarithmic frequency (6, 12, 18, ect., dB/octave).

When a test of a system shows the presence of any of the above, and any of the elements in that system could produce any of those features on its own (this includes the mic as well), then any attempt to separate the contributions of the components constitutes a guess. There is no general set of speaker behaviors that can be used to identify whether the speaker is contributing any of the above; the speaker could be causing all (i.e., the amp's linear frequency response is "flat") or none (speaker response is flat) of them. This is why I point out the futility of guessing.

If a person were interested in setting up a "profiling" system (actually a "modeling parameter-acquisition system" in the KPA's case) that could separate the amp from the cab, it could be accomplished. It would require accessing the electrical signal use to drive the speaker in addition to the signal from the mic.

[Gasp100]

Originally Posted by m~Dan

Hm... I don't have an amp here (and no room where I could profile without my neighbors freaking out because they probably think a UFO has just landed).

But I could try it with a preamp and different IRs instead of a real cab ... that should show how good it works too, right?

Does anyone have any concern over the loudness of the profiling session? I'm not too concerned now, because I will be relying heavily on others profiles. But if/when I profile, super loud UFO sounds don't sound like much fun. I mean, I use modelers because I can't turn an amp up in the first place. Now if I want to profile I have to get the amp into the sweet spot and than ET phones home?
How loud are we talking?
Is the loudness dependant on the volume of the amp you are profiling at the time?

[Jay Mitchell]

Originally Posted by m~Dan

Hm... I don't have an amp here (and no room where I could profile without my neighbors freaking out because they probably think a UFO has just landed).

But I could try it with a preamp and different IRs instead of a real cab ... that should show how good it works too, right?

How are you going to test the amp/cab combination, then? The experiment needs to be:

1. Profile amp/cab. One profile is enough.

2. Separate the "amp" portion.

3. Compare the following: 1) separated "amp"/clean power amp/physical cab to )2 physical amp/physical cab. If the KPA's guess is a good one, then the two should sound the same.

[Gasp100]

Originally Posted by mattball826

im no fan of any one device or method. its all here for our use if we want it. jay knows less about the kpa as anyone else do here commenting base on clips. call it marketing as you wish but is no different than claims other modeling designers make.

want scientific? modeling is not accurate as the real thing. it always has flaws and will never match analog 100%. kpa or axe or 11r or pod will have digital component to sound that analog amp does not. imo, some digital modeling method may be too perfect which why sounds synthetic to some users.

what "slays" is open to user interpretation. for me a traditional 50-100w head and 4x12 will destroy any amp model and frfr monitor in tone, cab response, dynamic feel.

kpa is not a modeler. it emulates a sound of the amp cab rig being sampled. some may like that or it may seem more real to them. others it will not.

Have you ever played the AxeFX II into a very high quality power amp into a great cab? Check out some of Mark Day's high gain clips and some of his comments about using the II with the new Atomic power amp into his cabs of choice.

[Jay Mitchell]

Originally Posted by Gasp100

Does anyone have any concern over the loudness of the profiling session?

It's interesting to me that this very significant issue has taken so long to appear.

Now if I want to profile I have to get the amp into the sweet spot and than ET phones home?

Uh-huh. Unless you want a profile of your amp being played well below its happy place. And if that's what you want, then you can just turn down the physical amp and play to your heart's content.

How loud are we talking?

How loud is the amp when it really starts singing? That's how loud we're talking.

Once you get really good at acquiring profiles - it's abundantly clear from user reports that this is a nontrivial process and can take a number of tries before you get it right - you'll only have to do the really loud ET-phone-home thing once.

[Gasp100]

Originally Posted by Jay Mitchell

Wrong. When I say "accurate" in this context, I am saying that it will sound substantially different than the "real deal."

What I meant to say is that the SPLIT of amp from it's originally profiled cab may not sound exactly like the "real deal" (I guess now we are talking about the amp itself, without a cab?). AND, I will not be completely accurate from a scientific point of view either. It will be an approximation... BUT, if I can pick up that "amp" which has been split from the original cab and slide in another cab which has been split from a completely different profile and it sounds "good" who gives a sh*t.
BTW, once it is all in the box do people really know how close this process is? I guess you would have to create two profiles of an amp using two seperate cabs. Then split and swap cabs. Then make your decision.
EDIT: I see Jay's response is more accurate. You always want to compare against the real deal (original amp plus cab of choice) to see how close it is.

[m~Dan]

Originally Posted by Jay Mitchell

How are you going to test the amp/cab combination, then? The experiment needs to be:

1. Profile amp/cab. One profile is enough.

2. Separate the "amp" portion.

3. Compare the following: 1) separated "amp"/clean power amp/physical cab to )2 physical amp/physical cab. If the KPA's guess is a good one, then the two should sound the same.

I have an amp/cab in rehearsel room. But I won't be there till next weekend.
I could try:

1. Profile my preamp with an IR.
2. Seperate the amp.
3. Compare the profile with preamp into the same IR I profiled.

That should basically give an idea how good it works, or am I completely wrong?

@Gasp: Like Jay says... the profiling can be very quiet. It depends on your preferred "ampvolumesoundsetting" I wouldn't do the profiling at home. But I wouldn't play a tube amp at a "comfortable" level at home too... so... I'm doing the profiling at rehearsal room, results are pretty good regarding the fact that I never ever miced a cab before. I'm placing the mic using in-ears to monitor.