Question for Cliff about the IR Technology and the future of Cab Simulation

Here was what Cliif stated earlier about the next generation of cab modeling. I don't think this will make into the 2 though:

" I've done tests using just a few excitation levels and, to my ears, haven't noticed any difference.
+1
I made a test with Nebula and came to the conclusion that the quality of the capture is the most important thing to get a good IR.
It's easy to find a static IR which sounds better than Nebula's IRs so...until somebody prove me that Nebula does a better job with IRs, today for me it doesn't.
(but Nebula is an amazing tool for other things)
 
"There are some little problems inherent the IR that let me feel that the amp and the cab are not "connected" together."

This is accounted for in the amp block (look at the low and high freq resonance parameters).

from the manual:

"LF RESONANCE, LF RES FREQ, LF RES Q – Guitar loudspeakers have a low-frequency resonance,
typically about 100 Hz. This typically shifts up slightly when the speaker is mounted in an enclosure. This
resonance causes an increase in the power amplifier response due to the finite output impedance of the
power amp. These parameters set the magnitude of the resonance and its center frequency and Q, or width.
HF RESONANCE, HF RES FREQ, HF RES Q – A loudspeaker voice-coil presents an inductive load to the
power amp at high frequencies. This inductive load, in conjunction with the output transformer capacitance,
creates a high-frequency resonance. This resonance causes an increase in the power amplifier response due to
the finite output impedance of the power amp. These parameters set the magnitude of the resonance, its
center frequency and Q, or width."

Also the cab size was added in the last update.

The problem with these parameters is that they're very difficult for most users to set accurately. You can adjust them by ear, but many of the parameters in the signal chain are interactive to some degree, so it's easy to get lost in a haze of options. I think this is one of the areas which most confuses users who find the Axe difficult to work with.

I think one thing that would help would be to have some way of having these values set automatically when you choose a cab (or adjust some relevant cab parameter, if there are any), at least for the factory cabs. The "correct" values vary from cab to cab, but there's no list (AFAIK) of what those values are and thus no way to apply them. In the real world, these values are not directly set by the amp designer; instead, the cab "talks" to the amp by loading it and they automatically "negotiate" the actual value.

I believe Jay or Cliff or one of the other cab experts has asserted that the sonic impact of these parameters is pretty minor, so it may not matter. I expect that a lot of users do like I do and simply leave these set to the defaults all the time. Tweakers* may use them as a final polishing step. But if the effect is so minor, that does beg the question of why they're exposed at all. In the absence of an auto-link feature, this means that most Axe users have the "wrong" values for these parameters and (however minor the impact) this will compromise how close they're able to get to the "real thing". The standard response here is "use your ears", but in this case at least there is a measurable "correct" setting, it's just not readily available to the player. That kind of stuff should be internalized where possible, IMO. Currently they're as likely to cause confusion as they are to enhance the player's experience.

As for cab modeling vs IRs, there's no question that IRs are more accurate than any model at a specific fixed point. The advantage of a model is not increased realism but increased flexibility - in terms of mic positions, cone distortion, and perhaps weird things like open-backed cab effects which are challenging for IR-based approaches. If you're not 100% happy with a particular IR, you can't just nudge the mic in a little closer, you have to switch IRs. The holy grail would be some sort of cross between IRs and modeling that would provide a reasonable approximation of what an IR hears, but with the greater flexibility of a model. It sounds like Cliff feels that the current system (1024-point IR + drive + room size, etc) is as close to that sweet spot as we can get with the current horsepower (since an IR-interpolation system would be highly memory intensive but not significantly better in results). I'm not remotely qualified to judge that myself, so I'll have to take his word for it. I'm glad he's constantly working to improve what he has.



*(you know, people who do 28 Adderalls before they sit down in front of their Axe)
 
Last edited:
Actually in the II the resonance parameters can significantly alter the tone. No doubt the amount of parameters can get overwhelming. A list of a values would be nice.
 
About the IR. Using test tones that goes from 0 to 20000 HZ seems a little bit strange for a really limited device like a guitar cab that usually goes from 60-80hz to 5k-6k. It's this approach that let me think that the technology is a little bit strange, guitars don't produce sine or pink noise sweeps.
I know it's a way to evaluate the cab frequencies spectrum but this is the reason why IMO it sounds a bit strange.

Are you saying that any IR captured using the sine sweep method is flawed?

Richard
 
Are you saying that any IR captured using the sine sweep method is flawed?

Richard

Absolutely not, maybe I've expressed myself bad.
I just think that this kind of technology is really at its infancy. It's like drum sampling. Back in the days the snare was just note, recorded at a certain velocity/strenght and then dropped/raised in volume. Look at where we are now.
That or the approach to physical modelling that the guys here already answered that right now would be just too heavy for actual processors.
I've just asked because I wanted to know how people with a lot more knowledge than me approached the thing.
It's just me, I become really geeky about physical modelling :)
A few months ago I have found a Piano Virtual Instrument called Moddart Pianoteq.
What an amazing piece of software!!
Have you ever tried it?
 
Absolutely not, maybe I've expressed myself bad.
I just think that this kind of technology is really at its infancy. It's like drum sampling. Back in the days the snare was just note, recorded at a certain velocity/strenght and then dropped/raised in volume. Look at where we are now.
That or the approach to physical modelling that the guys here already answered that right now would be just too heavy for actual processors.
I've just asked because I wanted to know how people with a lot more knowledge than me approached the thing.
It's just me, I become really geeky about physical modelling :)
A few months ago I have found a Piano Virtual Instrument called Moddart Pianoteq.
What an amazing piece of software!!
Have you ever tried it?

Oh okay.

I was commenting on what you said about being able to hear something undesirable when test tones were used to capture an IR. I believe you were talking about the typical frequency response of a guitar speaker vs. the test tone frequency range used to capture the IR.

Richard
 
could all this end up as having a law of ever dimminsing returns effect???

essentially going to great lengths to add another dozen parameters which don't offer anything in a truly meaningful way that couldn't be got extremely close to via the controls you have already...

and at the end of the day... some of these differences are so minute and so subtle, would they really impact your playing in a significant way?? and wouldn't they go largely un-noticed in the mix??

at what point would you consider modelling to be very very good, usable, does what it's supposed to, sounds and feels great...
or is engineering for the sake of it... going too far with to little gain... way to far... like right off the face of the Earth and headin' out to Uranus.. lol..
 
I just think that this kind of technology is really at its infancy. It's like drum sampling. Back in the days the snare was just note, recorded at a certain velocity/strenght and then dropped/raised in volume. Look at where we are now.
Infancy? The *maths* have existed for quite some time (early 20th century?). *Practical* applications in regard to musical application/use are more recent; I believe Sony was one of the first with an *affordable* rack unit (circa $5k US I think??? originally, released.. '99?). Soonish after, IR capable plug-ins popped up. Acoustic Mirror was my first experience (circa.. late 2000?), and I was in on early testing of Waves offering (I was VERY vocal/involved). The science is solid, and as processing ability/capability increases, we are afforded *better* operational benefit. However, please do not take that to mean that what we have now as being.. lacking. I am quite sure Cliff (and other math/engineer types, especially?) would provide deeper explanations and examples.

The comparison to the drum samples thing might actually be appropriate, but only glancingly IMO. Early sample based drum offerings (Drumkit From Hell being the *first* truly successful offering, circa 2001/2002?) had limited numbers of samples due to processing limitations, which is/was intrinsically linked with software *capabilities*. As processing power/capability increased (affordable), so software capabilities followed. DFH, in fact, had multi-samples for each kit piece. Though the product was 2 CD-ROMs, one only contained recorded pieces. The similarity? Software capability had to coincide with hardware capability. Currently, hardware capability allows for current "in use" technologies to be developed (nearly) as far as those technologies allow. Consider: would a rack of servers provide for a *better* sounding AxeFx using the current "software technologies" already in use? Or will new science/maths and thus new software be required? Is it more a matter of refinement (algorithms, programming, etc)?

As far as "why sweeps": The reading material is readily available.

Why use a "full range" sweep for a guitar speaker? It comes down to math I believe. Above 20k is another debate entirely, and that table is long worn, and likely to have some blood stains, lol. As far as "static" versus "reactive," this would be pure speculation on anyone's part, save Cliff (or anyone he has related his design theories with). I guess we could compare this, once again, with those drum sample libraries, in that each *bit* (punny? lol) is more akin to photo than motion picture, which, to carry the comparative further, when compared with "live" would be.. "missing that something." Could modeling technologies be integrated to account for this? Are solely IR/convolution based technologies too static for something so dynamic? On the latter, of course. An acoustic space will not have *reactive variance(s)* that are necessary to consider, therefor allowing the *static* nature of convolution/IR technologies to provide quite well. One could theorize that... say, combine three technologies: one, IR capture; two, an adequate number of varied, dynamically representative IR captures to properly represent the character of a time/source dynamic/timber variant (speaker), similar to the mega-layer + velocity piano/drum/etc libraries; three, modeling technology, evolved (incorporate those "IR samples," plus a "variable predictive variance and connectivity" aspect) , so as to incorporate the other two technologies, and thus provide the *missing analog* (analog here as comparative (sic), not electronic component, per se). IOW, modeling is missing.. something, and sampling can provide it, to some degree, and vice versa for sampling. Interestingly, Digidesign's "Velvet" utilizes these technologies (to a certain extent). I am unable to elaborate, but suffice it to say that my very first exposure to this was.. well appreciated.

Summarizing: I believe Cliff is working toward the same goal we all wish for. I tried an Axe Ultra, and found it lacking in such a (vitally) minute amount that given a choice between "stable of amps + cabs" and it, I had to (very grudgingly) let the Axe go. Axe Fx II, however, clearly... VERY clearly.. demonstrates the impact of improved hardware technology, a very solid foundation (Axe Stand., Ultra + incredible user base), and time spent on refinements. Cliff has explained (some?) of the major advances in regard to the IR + modeling technologies being utilized. Will the next step be one similar to what I described (ir + ir "layers+layered velocities" + incorporative, variable predictive modeling")? Or will it be something we are not aware of currently? IMO, it is close enough for many. For some, using a power amp + cab + mic is still preferred. EMWV (everyone's milage will vary.. to some extent :) ).

I could be spilling effluvia via my backside though...
 
could all this end up as having a law of ever dimminsing returns effect???

essentially going to great lengths to add another dozen parameters which don't offer anything in a truly meaningful way that couldn't be got extremely close to via the controls you have already...

and at the end of the day... some of these differences are so minute and so subtle, would they really impact your playing in a significant way?? and wouldn't they go largely un-noticed in the mix??

at what point would you consider modelling to be very very good, usable, does what it's supposed to, sounds and feels great...
or is engineering for the sake of it... going too far with to little gain... way to far... like right off the face of the Earth and headin' out to Uranus.. lol..

I agree with you on the fact that sometimes we focus on really subtle things, but on the other side I have to say that I like this approach.
I mean, you know that those parameters are there, it's up you if you want to tweak them or leave it at factory settings. And there are already, I think over 1000 parameters to change in the Axe, a dozen more will not be a problem :)

Infancy? The *maths* have existed for quite some time (early 20th century?). *Practical* applications in regard to musical application/use are more recent...

Nikki, great to know that you now feel the II improved in those areas you were worried about.

When I say in its "infancy" it's because the technology isn't moved a lot from "one IR" at a time.
And other than Velvet from Digidesign there is also TruePianos that uses modelling and sampling.

What I wanted to say in this post is that, we are moving to the full circle of "modelling" a guitar based chain, and just like the huge improvement between Kemper, Pod (static) and Axe Fx (dynamic) there is "room" to improve the latest and most important part of the chain IMHO, the cab.
And I really like the fact that the guys are working and studying on subjects to improve it, as it seems from the new "Cab Size" parameter.

But keep in mind that this is not an attack because I feel there is something lacking the Axe, a lot of people become really defensive on some subjects, I don't know why, it's just a thought that I had on a subject that I wanted to speak about it and have it clarified on some aspects.

Paolo
 
Actually in the II the resonance parameters can significantly alter the tone. No doubt the amount of parameters can get overwhelming. A list of a values would be nice.

I was having a lot of trouble with the Matchless on the II until I started changing the resonance parameters. Not exactly sure what I did but all of a sudden there was magic!
 
Back
Top Bottom