Darkglass Anagram - NAM profile player & multieffects unit

[Dom Mcsweeney]

Well, modellers at some point all ended up adopting IR technology.
I see it happening very similarly for NAM.

And, besides, no one is forcing a company to use the latest developments of any open source repo - that would be madness.
Just stay with what works.

 

[ChordInversion]

I’ve seen a few people staunchly against the TMP getting captures. I see the pros and cons but ultimately I think it would attract more users to the platform because it’s a feature people seem to love. I’m personally indifferent but I think it would open up their customer base a lot. They currently do updates every 6 months or so and add a couple of amps and pedals. It’s not a crazy output at the moment so it’s not like adding captures is going to insanely slow things down even more.

It almost certainly slow things significantly, though I won't assert a threshold for "insanely." :) If their updates are so labor-intensive (or their team is so small) that they can only manage a few feature and amp updates every 6 months, an additional parallel development roadmap is absolutely going to slow it further. Finite resources mean trade offs.

While I am making the case for why I don't consider it to be desirable on modelers (for the manufacturer), I personally don't care if any modeler adds capture, so long as: 1. They prove me wrong and it doesn't slow down feature enhancement, amp/effect model addition, etc., and 2. It doesn't hit my interface at all unless I'm intentionally using it.

It doesn’t take away it only adds imo. The QC seems to be doing fine with captures and models from day one and I can kind of see this path happening for line6 and possibly fender at some point.

The QC has been stating and not delivering (or delivering very, very late) from day one, too. I wouldn't use them as the exemplars of engineering effort planning and execution.

As far as NAM and being at the mercy of open source dev. I can see them forking NAM and doing their own thing to stay in their own lane away from constant nam development. In theory if they test things out and it works then the future dev is upto them, they’re not at the mercy of an evolving open source platform.

It's plausible, yes. But that's a whole different parallel development roadmap (once you fork, you're owning a new product - but still wanting to maintain some connection to the trunk for reasons I mention below), and it comes with a lot of pitfalls. Will you make sure players would be able to use all the other NAM captures out there? There would have to be extensive additional testing to accomplish that to be sure - on an ongoing basis. And if you abandon the rest of the NAM world, how much have you added? Few people actually make captures, so much of the point is the ability to buy/sell/exchange them.

It's all pretty interesting. Then there's the marketing angle. On the one hand, more is more. On the other hand, you counterfeit yourself a bit, and maybe a lot. If you add captures and people like it better than your modeling in the ultimate A/B test (on the same device), your flagship product just became an extremely expensive Tonex with some extra effects. The branding people probably hate the idea. They're trying to sell the notion that "our brand has secret sauce tied to our skills (and in Fender's case, history) that cannot be matched." To go commodity capture implies that the secret sauce doesn't really matter as much as they say it does.

 

[ChordInversion]

Well, modellers at some point all ended up adopting IR technology.
I see it happening very similarly for NAM.

Could be. I think some people like modeling,a nd the use cases and player experiences are fundamentally different. It's the "modeled knobs vs pulling another capture" difference. There's no right answer, IMO, but it's not a rift you can reconcile, IMO.

And, besides, no one is forcing a company to use the latest developments of any open source repo - that would be madness.
Just stay with what works.

Sort of. There's the matter of maintaining compatibility with all the captures out there. That would require at least some degree of keeping current as the spec changes.

 

[2dor]

It almost certainly slow things significantly, though I won't assert a threshold for "insanely." :) If their updates are so labor-intensive (or their team is so small) that they can only manage a few feature and amp updates every 6 months, an additional parallel development roadmap is absolutely going to slow it further. Finite resources mean trade offs.

While I am making the case for why I don't consider it to be desirable on modelers (for the manufacturer), I personally don't care if any modeler adds capture, so long as: 1. They prove me wrong and it doesn't slow down feature enhancement, amp/effect model addition, etc., and 2. It doesn't hit my interface at all unless I'm intentionally using it.


The QC has been stating and not delivering (or delivering very, very late) from day one, too. I wouldn't use them as the exemplars of engineering effort planning and execution.


It's plausible, yes. But that's a whole different parallel development roadmap (once you fork, you're owning a new product - but still wanting to maintain some connection to the trunk for reasons I mention below), and it comes with a lot of pitfalls. Will you make sure players would be able to use all the other NAM captures out there? There would have to be extensive additional testing to accomplish that to be sure - on an ongoing basis. And if you abandon the rest of the NAM world, how much have you added? Few people actually make captures, so much of the point is the ability to buy/sell/exchange them.

It's all pretty interesting. Then there's the marketing angle. On the one hand, more is more. On the other hand, you counterfeit yourself a bit, and maybe a lot. If you add captures and people like it better than your modeling in the ultimate A/B test (on the same device), your flagship product just became an extremely expensive Tonex with some extra effects. The branding people probably hate the idea. They're trying to sell the notion that our brand has secret sauce tied to our skills (and in Fender's case, history) that cannot be matched. To go commodity capture implies that the secret sauce doesn't really matter as much as they say it does.

So in the end what are you trying to get across?

I mean why would we, the consumer, care for what it costs vendors bringing profile playback abilities into their units? I'm pretty sure they're all big boys & can manage their product roadmaps / portfolios at the end of the day.

 

[Lysander]

Yes, but the integration into their proprietary code is not done. And it's a significant effort. And NAM is under active development. What it takes to implement it will absolutely change. To adopt it means a perpetual development roadmap. Integration of open source standards always does. It has to.

This is 100% correct - and often a neglected point when NAM hardware is discussed. NAM is a (very!) moving target right now, and it will continue to be for the foreseeable future. Which is not a problem for plugins, but it is definitely a problem for hardware with long lifespans. Even @Digital Igloo chimed in explaining the same after a couple questions about NAM integration in Helix.

When NAM hits v1.0.0 a couple years from now, will captures work with existing hardware? Who knows.

 

[ChordInversion]

So in the end what are you trying to get across?

That it would involve a considerable trade off, and that we shouldn't expect that the high end modeler manufacturers can (or would even want to) just drop it in.

I mean why would we, the consumer, care for what it costs vendors bringing profile playback abilities into their units? I'm pretty sure they're all big boys & can manage their product roadmaps / portfolios at the end of the day.

Because of the trade off. It simply has to impact the rate of updates and feature enhancements that aren't related to capture that will make it into those products. If you have a modeler and want it to keep adding models and features (to the modeling functionality) as quickly as they can, adding NAM block support would work against that goal - and on an ongoing basis. It has to.

 

[2dor]

This is 100% correct - and often a neglected point when NAM hardware is discussed. NAM is a (very!) moving target right now, and it will continue to be so for the foreseeable future. Which is not a problem for plugins, but it is definitely a problem for hardware with long lifespans. Even @Digital Igloo chimed in explaining the same after a couple questions about NAM integration in Helix.

When NAM hits v1.0.0 a couple years from now, will captures work with existing hardware? Who knows.

There are 2 parts to NAM:

- training
- inference (playback)

Most enhancements were brought into the training piece. Profiles from early versions are pretty much able to run in the latest NAM player - it's just that newer profiles might have additional metadata (SEND and RETURN) for the input level calibration. Aside from that, that's pretty much the only major development really. At the end of the day, a NAM profile can be opened in a TXT editor and it's a bunch of metadata and JSON weights for the WaveNet.

Steve's been pretty adamant about having a fixed tiered order for the profile types and that hasn't changed in +2 years since NAM started getting popular.

As Nathan pointed out, there's nothing holding back devs from keying in their own limitations to what can & can't run. It'll still be WaveNet or LSTM at the end of the day.

 

[laxu]

Could be. I think some people like modeling,a nd the use cases and player experiences are fundamentally different. It's the "modeled knobs vs pulling another capture" difference. There's no right answer, IMO, but it's not a rift you can reconcile, IMO.

I'm definitely on the side of modeling where the amp behaves like the real thing too, rather than just being a snapshot of it. There's a lot of nuances to be found on any amp that you lose with captures because they only represent a subset of settings.

On the flipside there's plenty of people who don't know what to do with the real amp so they will be happy scrolling through captures because it means they don't have to learn anything. "Sounds good, let's play".

Sort of. There's the matter of maintaining compatibility with all the captures out there. That would require at least some degree of keeping current as the spec changes.

There is versioning in the NAM files, so that can help manufacturers with support but of course the end user does not know any of that. To them it would be "This NAM capture does not load on my new modeler!!!!11" because it's some new version that is not supported, unless the UI tells that that is the case.

 

[[Nathan]]

It almost certainly slow things significantly, though I won't assert a threshold for "insanely." :) If their updates are so labor-intensive (or their team is so small) that they can only manage a few feature and amp updates every 6 months, an additional parallel development roadmap is absolutely going to slow it further. Finite resources mean trade offs.

While I am making the case for why I don't consider it to be desirable on modelers (for the manufacturer), I personally don't care if any modeler adds capture, so long as: 1. They prove me wrong and it doesn't slow down feature enhancement, amp/effect model addition, etc., and 2. It doesn't hit my interface at all unless I'm intentionally using it.


The QC has been stating and not delivering (or delivering very, very late) from day one, too. I wouldn't use them as the exemplars of engineering effort planning and execution.


It's plausible, yes. But that's a whole different parallel development roadmap (once you fork, you're owning a new product - but still wanting to maintain some connection to the trunk for reasons I mention below), and it comes with a lot of pitfalls. Will you make sure players would be able to use all the other NAM captures out there? There would have to be extensive additional testing to accomplish that to be sure - on an ongoing basis. And if you abandon the rest of the NAM world, how much have you added? Few people actually make captures, so much of the point is the ability to buy/sell/exchange them.

It's all pretty interesting. Then there's the marketing angle. On the one hand, more is more. On the other hand, you counterfeit yourself a bit, and maybe a lot. If you add captures and people like it better than your modeling in the ultimate A/B test (on the same device), your flagship product just became an extremely expensive Tonex with some extra effects. The branding people probably hate the idea. They're trying to sell the notion that "our brand has secret sauce tied to our skills (and in Fender's case, history) that cannot be matched." To go commodity capture implies that the secret sauce doesn't really matter as much as they say it does.

I say this in the most respectful way possible but it’s like your spirits have been crushed by your work history or something. At the end of the day Fender has resources, there are many many competent devs out there. And at the top of the totem pole even people like Steve, Mike and other devs in the field would surely lend a hand or have contract rates, it’s all doable if they wanted to.

The TMP is a computer and it just needs to integrate something that’s already built, working and has been integrated by other things already (Darkglass, dimehead and many software applications). It’s really not that hard of a task in the grand scheme if they wanted to do it. You can point out a million sceptical reasons why a new building shouldn’t be built, or you can put in the work and just get it done.

I’m not even advocating for the TMP to get captures but if they come I’d use them and also be a contributor. I just think of all the other software/hardware hurdles that have been conquered, and adopting NAM into a line6 or TMP is pretty low on the totem pole stuff in the grand scheme.

 

[Lysander]

Most enhancements were brought into the training piece. Profiles from early versions are pretty much able to run in the latest NAM player - it's just that they might have additional metadata (SEND and RETURN) for the input level calibration.

I understand how NAM works, but, again: you're talking about the past, and past performance is not indicative of future results . NAM is 5 years old, and only had significant work put into it by 2022; we've had major updates happening in just three years, including parametric models and multiple improvements in the training process.

Point is, NAM is developed as a desktop piece of software... because it is. It involves completely different considerations from hardware-specific algos.

The TMP is a computer and it just needs to integrate something that’s already built, working and has been integrated by other things already (Darkglass, dimehead and many software applications). It’s really not that hard of a task in the grand scheme if they wanted to do it.

Nit: not all "computers" are built the same. Ask Neural DSP and their three-year long journey to port desktop plugins to DSP-based hardware.

 

[Sedaxel]

I often see the example of IRs, comparing it to captures. I´m not an IT expert, but I guess they´re quite different things. IRs use FFT, convolution/deconvolution... (I don´t know the details, honestly). AFAIK it was just using existing tools. And IRs are made in standard audio files.

Capturing, on the other hand, needs especific code to be run. Code that is in continuous development. Captures are not presented in standard files, but in especific ones for each capturing platform.

(all of the above can be corrected savagely... as I don´t really know anything about details)

If I put myself in the skin of a pedals brand, the last thing I want is to dedicate resources in something that... is just going to be exactly the same as the rest of manufacturers. Why would I bother if I´m not getting any competitive edge? I mean, Fractal invests to have the best modelling out there. This will guarantee them a technical advantage, and sales based on it. But... if all the brands sound as the last version of NAM, why would I buy one or the other? I´ll just go for the cheapest, so expensive ones would quit (you´re similar to the cheapest, my friend... can´t buy your thing).

I don´t see NAM being implemented in all pedals. I certainly see capturing being a comodity, but something propipetary, so brands can invest, research, and get the edge over the competition.

Or at least, a few with NAM, and the top tiers with something propietary.

Or maybe it´ll vary with coming years... who knows!

 

[Dom Mcsweeney]

Everything is pretty much backwards compatible currently.
And, going forward, could be too.
If not, it can be worked out easily - Steve and anyone involved are only an email away from one another.
I can't imagine a point where 'oh dear, suddenly everything is not working'.
Outside of that - probably the users who don't have amps of their own to train, are going to want a future where the 'community uploads' are still accessible - and even that can be worked out easily via email if one website or server needs to take over another etc.
I say 'email' , but obviously other steps would be required, I'm just simply pointing out everything is do-able if communication happens... unlike 'guessing negatively that imaginary problems are certain'.

 

[ian_dissonance]

I say this in the most respectful way possible but it’s like your spirits have been crushed by your work history or something. At the end of the day Fender has resources, there are many many competent devs out there. And at the top of the totem pole even people like Steve, Mike and other devs in the field would surely lend a hand or have contract rates, it’s all doable if they wanted to.

The TMP is a computer and it just needs to integrate something that’s already built, working and has been integrated by other things already (Darkglass, dimehead and many software applications). It’s really not that hard of a task in the grand scheme if they wanted to do it. You can point out a million sceptical reasons why a new building shouldn’t be built, or you can put in the work and just get it done.

I’m not even advocating for the TMP to get captures but if they come I’d use them and also be a contributor. I just think of all the other software/hardware hurdles that have been conquered, and adopting NAM into a line6 or TMP is pretty low on the totem pole stuff in the grand scheme.

Is it a computer or is it a computer controlled DSP? I honestly don’t know, but they’re not 1:1 as far as programming and software.

 

[Mike Oliphant]

NAM is a (very!) moving target right now, and it will continue to be for the foreseeable future.

While I can't predict the future, NAM playback has actually been quite static for years now. Both the player and the trainer have been improved, but there hasn't been a breaking change in the file format for quite some time.

 

[ChordInversion]

I say this in the most respectful way possible but it’s like your spirits have been crushed by your work history or something. At the end of the day Fender has resources, there are many many competent devs out there. And at the top of the totem pole even people like Steve, Mike and other devs in the field would surely lend a hand or have contract rates, it’s all doable if they wanted to.

Heh. I get that. My work history hasn't crushed me at all. It's made me more and more effective. I'm just saying resources are finite. Some people see open source and think that means it's free to implement. It very much isn't, and the expense is ongoing, not one-time. Wishing otherwise doesn't make it so.

The TMP is a computer and it just needs to integrate something that’s already built, working and has been integrated by other things already (Darkglass, dimehead and many software applications). It’s really not that hard of a task in the grand scheme if they wanted to do it. You can point out a million sceptical reasons why a new building shouldn’t be built, or you can put in the work and just get it done.

Yeah, but some buildings get built and some don't, and gumption isn't the reason. Whether they make sense to the builder is the reason. It has to be worth the cost to the builder. I question whether it is in this case.

I’m not even advocating for the TMP to get captures but if they come I’d use them and also be a contributor. I just think of all the other software/hardware hurdles that have been conquered, and adopting NAM into a line6 or TMP is pretty low on the totem pole stuff in the grand scheme.

Kinda/sorta. I think it involves a lot more ongoing effort (which would take the place of other ongoing effort) than many folks are acknowledging.

 

[2dor]

and here we are on the Darkglass Anagram thread

 

[Lysander]

and here we are on the Darkglass Anagram thread

We really need more bassists yelling at clouds over here

 

[MirrorProfiles]

Well, modellers at some point all ended up adopting IR technology.
I see it happening very similarly for NAM.

Haven’t IR’s outdated modellers by a long time? Pretty sure the Pod V1 was using IR’s and so has basically everything since. I don’t think the ability for users to load their own was really a technical limitation (because modellers already used IR’s) as much as figuring out an elegant way for users to import and manage them.

TLDR is that IR’s are surely easier to support because the tech is in there already. Adding NAM support to something that isn’t using something similar already is much trickier.

 

[Lysander]

Haven’t IR’s outdated modellers by a long time? Pretty sure the Pod V1 was using IR’s and so has basically everything since.

No, Pod's A.I.R technology was built around EQs.

 

[Sedaxel]

No, Pod's A.I.R technology was built around EQs.

AFAIK, Zoom G5n cabs were IRs. That´s the first use of them I´m aware of. Maybe there were others before...