Kemper Profiler MK 2

[Chocol8]

This is sureal. Based on the excellent and detailed explanation of why null testing is a flawed approach to make the determination, we learn the only test that can provide proof is getting real guitarists to play through the test subject gear and deciding if it pleases them in all the multiple and nuanced ways a guitar amp does.

Yes

And then you will need a large sample size of those guitarist to log their approval or disapproval.

No.

The experiences of most of those guitarists don't matter. If a Kemper profile of a Deluxe Reverb sounds good for you, it sounds good for you. The fact that it can't handle someone else's Mesa dual rec and 7 string to their liking has no bearing on your experience.

Why do we need a one size fits all scoring system or ranking system? Its a fool's errand.

 

[Monkey Man]

There is hope!

 

[[Nathan]]

Kemper is either botting their videos or someone is botting it for them. There's a bunch of reasons why it's obvious to me but the comment to view ratio is really the dead giveaway. If you think their views are organic, 100% not.

JNC if you're lurking I give you my blessing to make a video on this

 

[Sascha Franck]

Liquid profiling is a bit crap

I actually think it's most excellent as it allows deeper tweaking of an otherwise static capture. Kinda like Genome's NAM block allows for vastly better tweaking than the OG NAM player or Tonex.

 

[Sascha Franck]

I can produce an even longer list than yours of theoretical shortcomings of integrated LUFS measurement of null test results.

What's missing though, is evidence that a simple integrated LUFS measurement of a null test discrepancy gives incorrect or misleading conclusions in a real-world comparison of guitar amp captures. The closest I've seen are contrived cases where the null test is intentionally performed incorrectly in order to introduce a LUFS difference that you can't hear.

Properly done null tests with integrated LUFS measurements that I've seen will pass the smell test of confirmation by listening, so all I'm saying is: I find it hard to conclude that there is a "need" for a more complex measurement.

I generally agree, but still, LUFS tests can't be "trusted", so to say.
Even while LUFS measurings are frequency weighted, you could still "cheat" things.

While absolutely oversimplified, let's just assume this scenario: An amp capture would only be inaccurate somewhere in the higher register, let's assume there'd be less highs than in the original. The null test would reveal whatever -XYZdB numbers. Now, if you raised some lower end, you could compensate so much that the numbers looked better. That would work regardless of LUFS being frequency weighted.
IOW: Two completely different signals could lead to the same LUFS values.

Now, as I'm not in the "authentic tone over everything" camp at all, I can't be bothered to examine things much further, but it actually seems to be like the tests Leo Gibson is doing are pretty much in line with whatever takeaways from blind tests. IOW, I kinda agree with you that they might be decent enough for the subject at hand.

Anyhow, for a more accurate nulltest one could of course split the signals measured into multiple bands and do completely independent measurings of those bands. Not sure myself whether it'd be worth the effort, though. But then, as said, I'm not a tone corksniffer by any means (and still considering to purchase a KPA Stage Mk1 one day).

 

[Lysander]

Now, as I'm not in the "authentic tone over everything" camp at all, I can't be bothered to examine things much further, but it actually seems to be like the tests Leo Gibson is doing are pretty much in line with whatever takeaways from blind tests.

FWIW, my gripe when people like Gibson uses loudness weighting to put numbers on null tests is not so much the result, but that they present it as scientific.

From a "science" perspective, LUFS have no business being used to quantify null test results. Useful, still? Sure.

 

[Sascha Franck]

From a "science" perspective, LUFS have no business being used to quantify null test results.

While I generally agree, they still kind of have. It's just not the science we need to evaluate the accuracy of captures as it treats all frequency areas equally (completely regardless whether LUFS is representing a weighted spectrum already), whereas in daily life (and in terms of perception) a large part of the frequency spectrum is just not relevant.
In case a capture is way off, say, below 40Hz, we may not even notice, but it might look miserable in a nulltest. As said, for more scientific validity, the signals needed to be split into "some" bands.

Yet, regardless of what we may think of Leo's tests - they correlate pretty fine with the results of whatever ABX tests.

 

[Lysander]

Yet, regardless of what we may think of Leo's tests - they correlate pretty fine with the results of whatever ABX tests.

Again: fine. Just don't call it scientific.

Measuring LUFS on null deltas for the lulz is all fine and dandy. The problem is when you start stacking profilers based on that number alone, because it doesn't measure profiling accuracy. Again, using poor Leo as an example - this is just plain wrong...

...and this is downright horseshit.

Anyone suggesting that, f.ex, NAM is 7% more accurate than ToneX because of a LUFS null measurement is misleading people - and that's being charitable.

 

[crispsandwich]

Again: fine. Just don't call it scientific.

Measuring LUFS on null deltas for the lulz is all fine and dandy. The problem is when you start stacking profilers based on that measurement alone, because it doesn't measure profiling accuracy. Again, using poor Leo as an example - this is just plain wrong...

View attachment 59620

...and this is downright horseshit.

View attachment 59621

Anyone suggesting that, f.ex, NAM is 7% more accurate than ToneX because of a LUFS null measurement is misleading people - and that's being charitable.

Leo is a tit. His null tests are shite. He's a decent player, he should focus on making music.

 

[Lysander]

Leo is a tit. His null tests are shite. He's a decent player, he should focus on making music.

I feel bad about focusing on Gibson because he's far from the only YouTuber doing this - but he's certainly the most prominent one.

 

[crispsandwich]

I feel bad about focusing on Gibson because he's far from the only YouTuber doing this - but he's certainly the most prominent one.

I think it's the fact he presents himself as some kind of authority on the subject. I could record a fart and do some 'tests' with it, and it would be just as 'valid' as the nonsense he spouts.

 

[Lysander]

I think it's the fact he presents himself as some kind of authority on the subject.

Yeah, that's fair - he does it a lot. See his (ignorant) takes on AI as well, which had me ranting about just this week.

 

[Jarick]

Kemper is either botting their videos or someone is botting it for them. There's a bunch of reasons why it's obvious to me but the comment to view ratio is really the dead giveaway. If you think their views are organic, 100% not.

View attachment 59614 View attachment 59615 View attachment 59616

JNC if you're lurking I give you my blessing to make a video on this

Yeah that is not normal viewing patterns at all. I've subbed to this channel for a while and they typically only get 2-3k views per video. Something is off there.

 

[DrewJD82]

I have a very basic understanding of LUFS from what I’ve learned about mixing and when I started seeing it mentioned in these comparisons/tests I scratched my head a bit and assumed my understanding of them wasn’t thorough enough to see how they could be used in that application legitimately. Glad I wasn’t off the mark.

 

[Orvillain]

A null test with LUFS works well for linear time-invariant systems (LTI).

• EQ plugins
• convolution reverbs
• cabinet IRs
• mic preamps below saturation

Because those systems obey superposition. If two LTI systems null closely then they are functionally the same device.

A real tube amp is:
Nonlinear + dynamic + memory-dependent

It is a state machine, not a transfer function. The output depends on:

• instantaneous amplitude
• recent past amplitude (sag, bias shift)
• spectral content
• note envelope
• playing articulation

So when you do a null test, you are only testing the behaviour of the amp under one exact trajectory through its state space.

That’s the core flaw, and it really wouldn't matter too much if we were using LUFS, dBfs, dBu, or any other weighted metering mechanism.

LUFS in particular though, assumes that differences are primarily amplitude-distribution differences. But guitar amp differences are usually:

• dynamic compression curves
• harmonic generation vs level
• pick attack transient handling
• bias recovery timing
• intermodulation distortion

Those do not strongly affect integrated loudness. This is why two amps can:

• feel completely different to play
• sit differently in a mix
• respond differently to palm-mutes

…yet null to -35 dB and have almost identical LUFS residuals. Because the perceptual difference lives in time-structure, not energy.

This isn't an ears versus measurements issue. This is a misunderstanding of DSP and amplification physics issue.

 

[Monkey Man]

I'm dumbfounded as to how LUFS could have even entered the discussion.

I mean, how many ºC do you weigh? Personally I'm 76ºF. Gotta lose a few hectares, obviously.

 

[Lysander]

I'm dumbfounded as to how LUFS could have even entered the discussion.

The reason is because it's there. Most, if not all, audio plugins give you LUFS meters because, well, having things compensated by loudness is usually pretty nice. When you have a hammer, everything looks like a nail.

But if you're trying to quantify null deltas in a "scientific" way, using k-weighted LUFS measurements is actually less accurate - it biases results by how human ears perceive volume.

 

[2dor]

I'm dumbfounded as to how LUFS could have even entered the discussion.

I mean, how many ºC do you weigh? Personally I'm 76ºF. Gotta lose a few hectares, obviously.

Well, they do paint part of the picture because you're looking at, what's ideally, 2 identical signals with the phase flipped over the course of an X seconds / minute sample.
The integrated loudness should give you a decent enough idea of how close the 2 signals sound with respect to each other.

I'm not saying his work is bullet proof but it's still a decent baseline; whether I like the guy or not, that's a different story.

Sure, there's stuff which can be improved upon but we only need 1 instance of that so folks can see how it's done & I'm sure we'll see a shift from the LUFS-I method to improved workflows.

 

[OneEng]

Not sure this is actually the case. Touch screens actually make it very easy and intuitive to control modellers. It's so much easier to see a parameter and tweak on screen rather than look for the knob or button that does something.

In that regard, I think the worst offender is the Axe FXIII, which is really difficult to manually programme. They have a damn good editor though, so that's a huge plus in its favour.

As a live gigging guy, I never use the Kemper interface to adjust things. This is done at home with a computer. Live, the only thing I ever touch is my foot controller.

For other users, I completely agree with you though.

The Kemper though... I really hate entering rig names for saved profiles, for example. And there's a lot of menu diving involved, which makes it very difficult to most users to do more than just scratch the surface of the beast.

Agree. I would really hate it if I didn't have a good PC interface to work with.

Looking at the Kemper V2, I just figure they thought they could save on development costs to maximise profits. I don't even find the advertisement of "xx more FX slots" as anything worthwhile when all the competition had that years ago.

I don't even use up the slots I already have. Not really even close.

It always becomes a "is this a good enough price that I can either sell it later, or eat the cost and keep it?" thing.
I see a lot of Kempers at around the 600 € mark on my local market. It's only going to be a race to the bottom to sell those.

I usually don't bite on old digital stuff because even though the utility value is high, they do tend to just keep losing value. Finding the right buyer is more challenging when everyone is going for the newer version.

Like I could get an Axe-Fx 3 for fairly reasonable prices used, but instead I pounced on the AM4 because it's the new thing and more in line with what I need.

This 100%!

That’s exactly how I designed my live rig to work in a ‘have my cake and eat it too’ situation, which is rare enough that I haven’t even gotten to try it yet. Contrary to internet forums, the overwhelming majority of local venues are not running stereo PA’s and even if they were, are not setup for an audience to receive a stereo mix in a way that would make it worthwhile, in most cases it would only cause a lacking of one side or the other.

It's enjoyable while practicing with headphones to get the EFX stereo in all its glory, but you are correct, this doesn't work live at all.

Agreed, that sounded spectacular. I have a mk1 rack unit. Really hoping we get to experience some improvements, though tbh I still think the v1 profiles hold up against pretty much everything else.

I strongly suspect that V2 will capture better than V1; however, after tweaking a bit, I don't think there will be a hill of beans difference. We will see.

Nice to see something from them. Waiting on the apples to apples V1 vs V2 comparison before further comment.

Aint that the truth!

Yep ... I don't care how it null tests against anything *other* than the same Profile done in the Legacy M1 way -vs- one done the new M2 way

And I do know null tests are very "limited".

But it will be a good / objective way to "hear" -and- "see" the differences ie:- not just do they subjectively sound and feel better [or worse (?)] but are they also objectively and mathematically better [or worse].

100%. Great example. One of Kemper's strengths is its ability to take someone else's profile and make it sound good on your own rig ..... easily.

Totally agree re: the Accuracy issue -this time- for Kemper V2 ... why ?

I have zero doubt C.K and the team know exactly how good NAM / Tonex are ..... QC to a slightly lesser extent ..... so to be taken more seriously they must get close-to-real-close to these other.

Where Kemper have a bit of a card up their sleeve is Liquid Profiling .... I used it for over 6 months for everything ... and yes .... with Mk1 there are only some 30+ or so Modeled Gain and Tone Stacks ..... was it perfect ... no .... was it like adjusting the G/B/M/T/P/ .... pretty bloody close and great feeling and responsive.

Static Captures on NAM / Tonex are awesome ..... but once you start adjusting the generic G/B/M/T in a Tonex / NAM / QC Capture <- i.m.h.o its just shitful.

See the following from Leo ... from 19m 54s <-> 24m 12s

QC V1 vs KPA MK1 Stock and Liquid Profile with Leo's Soldano ... in short

-> QC V1 Static -33.3 LUFS
-> KPA Legacy MK1 Static -30.9 LUFS

then

-> QC V1 Static with Setting Changes made via Generic QC Tone Stack - 28.3 LUFS
-> KPA Soldano Liquid Profile with Setting Changes - 33.5 LUFS

In short, once you move off the Stock Static Captured Settings

-> the QC V1 LUFS gets %15 worse
-> and the KPA Liquid Profile LUFS is %9 better
-> and head-to-head- the KPA Liquid Profile LUFS beats the QC V1 LUFS by nearly %16.

I am still thinking the old dog might have some new tricks up its sleeve once we see V2 Profiling (?)

Time will tell.

I didn't know it actually got more accurate when you manually tweak. Interesting. I knew I could make it SOUND better by tweaking .... never really caring if it was "accurate" or not, only that is sounded really good.

I don't see how liquid profiling is much better except it puts a tonestack model somewhere in between. Remember that the tone stack is potentially in the wrong place in relation to the real circuit, and potentially also uses the wrong values because it's afaik a generic one. Maybe it allows for intuitive enough control for someone used to the real amp.

However both solutions throw accuracy out of the window as you introduce more variables.

Null tests are probably a pretty bad metric for this too.

Agree.

Liquid profiling is a bit crap, honestly. If the profiles came out better there'd be less need for chasing down tonestack accuracy. I dont think it works particularly well either, it's kind of a botch fix. Quite telling that none of the better capture platforms really care for it

I think it makes it easier to tweak the tone to something desirable, but aside from that....

And if there is a big improvement, I honestly wouldn't be surprised if the Mk1 profiles created by the new capture method are extremely close to the Mk2 profiles. We saw from both Tonex and QC that there were improvements to be made on the same hardware platform using different capture processing tech. Kemper Mk1 vs Mk2 has the same DSP chip right?

I am thinking that MK1 profiles tweaked well will be a tough call to distinguish from V2 tweaked. I think that V2 will be better without tweaking.

I actually think it's most excellent as it allows deeper tweaking of an otherwise static capture. Kinda like Genome's NAM block allows for vastly better tweaking than the OG NAM player or Tonex.

Agree!

 

[OneEng]

I'm dumbfounded as to how LUFS could have even entered the discussion.

I mean, how many ºC do you weigh? Personally I'm 76ºF. Gotta lose a few hectares, obviously.

Isn't that a diaper brand?