Brian Wampler on capturing (a la NAM, Proteus)

Any suggesting this isn't the future has their head in the sand. The issue is once a capture/profile is done, how can it be edited. And this is where the next gen is going to go. By having a capture process combined with input meta data to describe amp topology and settings, there is the potential for a capture/profile paired with parameters mirroring the settings the real amp had and allowing one to adjust those mirroring how they would react in the actual amp. This absolutely requires an additional complexity to this style of devices which isn't there yet, but it will be. Kemper way back in the day proposed a tone stack library to pair to profiles to allow more realistic editing but the issue is getting those controls to default where the amp's were when the profile was taken.
 
Any suggesting this isn't the future has their head in the sand. The issue is once a capture/profile is done, how can it be edited. And this is where the next gen is going to go.

Eh, the way i see it, it's just another tool in the toolshed. Captures/profiles are very very good for certain use cases, but they're not perfect - nor they automatically sound better than the alternatives.

I think Brian is completely right about things like NAM and Proteus being industry-disruptive though: the entry cost for this technology is now literally zero.
 
The issue is once a capture/profile is done, how can it be edited. And this is where the next gen is going to go.

Seconded.
I guess one of the things we'll likely see is captures being kinda "broken up" into separate parts. Such as pre-drive, main drive and power amp drive (just as one possible example). All of these sections could then have their own tonestacks (and in/out levels, of course). You might then be able to combine them into a sort of macro, only exposing a traditional tone stack in case you don't want/need to go deeper.
Obviously, this also depends on the CPU demands of whatever capture playback devices.
 
I guess one of the things we'll likely see is captures being kinda "broken up" into separate parts. Such as pre-drive, main drive and power amp drive (just as one possible example).

But, once you go down enough that rabbit hole... you're essentially re-inventing modeling. This is how f.ex. NDSP models work under the hood. Line 6 does something similar with HX, and Positive Grid has been selling products where you can mix-n-match amplifier sections for a good while.

Frame_1554.png


The reality is, once you deviate from "i want this exact glorious amp sound on a box i can carry with me everywhere", capturing can become extremely limiting.
 
At 16:14 Brian suggests that free capture tech can be used to demo new gear before buying, which I have suggested myself in the past, but, people who buy pedals don't usually gravitate towards Modelers, DAWs, Plugins, etc. and then there's the fundamental difference between recorded sound vs hearing a real cab in the room, etc., so it might be more complicated and sound removed form the real product and it might cause more harm than good for sale figures.
The real product might be 'passed' just as quick as any shitty capture if the demo capture doesn't do it justice.

I also agree that adoption of this technology by big companies is not a matter of 'if' anymore but 'when'.
Just like IRs, this is going to be EVERYWHERE, modeling amps based on captures, capture loading pedals, etc.
 
I don't think it will be necessary to split anything. Give it a few years and hardware / AI accelerators will catch up and allow for full training, basically making perfectly profiled full models of amps. Now, that's going to be disruptive for amp makers, as it is an identically copy that invalidates the product.
What I would like to see is NAM model that would make use of community resources, kind of like project SETI or Prime95, where a distributed model would allow thousands of users to train a complex amp model.
 
Just like IRs, this is going to be EVERYWHERE, modeling amps based on captures, capture loading pedals, etc.

I don't know if this will ever happen, but if one capturing product comes out victorious and we end up with a de-facto standard then yes, i can totally see captures going the way of IRs.

NAM is likely the best candidate. Becoming an industry standard comes with drawbacks though; namely, the fact that protocols evolve (much) slower than platforms, so improvements on the NAM core would be quite difficult to materialize.
 
But, once you go down enough that rabbit hole... you're essentially re-inventing modeling.

Well, yes, maybe sort of - but I'm actually not asking for super detailed puzzle pieces. Really just something like a drive, preamp and poweramp that I (and others) could a) capture myself and b) combine myself and then treat it as a single block with idealized controls.
 
Just like IRs, this is going to be EVERYWHERE, modeling amps based on captures, capture loading pedals, etc.

Yeah, defenitely.
The really interesting thing possibly being what @Lysander already mentioned; will there be a sort of standard or will people go their own proprietary route? Could as well be a mixture of both - just as what we see with samples and IRs. The main files (in that case audio files) are more or less standardized, but the playback engines might add proprietary stuff and come with an additional, proprietary format, too - such as what big sample library vendors are doing. Or the folks into dynamic IRs.
 
I think “captures” is ultimately limited by the pain in the ass it is to demo them to find the right capture for your preferred guitar(s) and playback method, file management, lack of tweak ability, and the sorta “pro capture” after-market purchase nature of them that they naturally lean into.

Modeling simply and effectively eliminates all of that.

That said, the real breakthrough of the tech will be when the capture process effectively creates a model. So during the capture process you tweak and twiddle the knobs to some degree and the capture notes that info, and/or is then able to fill in the gaps (for the lack of a better word) effectively creating a full model rather than a snapshot. That has the potential to be really disruptive.
 
Last edited:
Having a modeled EQ that you can put before or after the capture with simulations of common tone stacks would go a looooooooooong way toward mitigating the controls issue, I think. If I could put an EQ that mimicked a marshall-style tone stack after a capture, I'm pretty sure it would scratch the itch. The ToneX EQ works pretty well so that I haven't really missed it that much but it shouldn't be that hard to approach from this angle- if it works as I suspect it could.
 
Having a modeled EQ that you can put before or after the capture with simulations of common tone stacks would go a looooooooooong way toward mitigating the controls issue, I think. If I could put an EQ that mimicked a marshall-style tone stack after a capture, I'm pretty sure it would scratch the itch.
Yes, I also think a hybrid approach is going to appear first, especially for high gain amps (Preamps).
For NMV amps that rely on poweramp and power supply voltage sag for their distortion like the Trainwreck Express or an early Plexi, captures will not cut it no matter how we twist it.
 
For me, the main appeal of capturing would be that I could simply combine multiple things into one capture. Sure, I wouldn't be able to access individual parameters within the source's "items" anymore - but I often don't do that anyway. For instance, I have a handful of drive/amp combinations that I never need to adjust from the ground. Altering the overall gain within what I'd need is easily possible using input trim and for adjusting things to suit a context I'm most often fine with a post EQ.
Hence, in case I'd be able to split things up into the 3 segments proposed above, I'd likely have more than enough combination variety and access points than I ever needed.
 
Give it a few years and hardware / AI accelerators will catch up and allow for full training, basically making perfectly profiled full models of amps.
Unless someone figures out a super fast way to capture the output of the amps, that's going to be far more of a limiting factor than how long it takes for the neural network to process everything. To account for all setting combinations would require a ton of captures even on a simple 1 channel amp.

I feel Fractal at least is already quite close to having their tech match to their real reference amps while also working like those amps. For me it would be more about Fractal managing to cram their tech into something smaller and cheaper than the FM3 that does just amp/cab sims. I'd take that any day over the capture stuff. Same would apply to Line6, I can't wait to see whatever their next gen "HX Stomp" will be.

Meanwhile the market will most likely get flooded by cheap capture player type pedals. Even if many of them won't have a library of tones to choose from, just being cheap and good sounding is enough. I wouldn't be even surprised if like a next gen Boss Katana has that sort of tech built into it.
 
Back
Top