Login

Slate Digital VCC N**e emu soundz real Gooooood!!

If you need to write about anything else please do it here..

Re: Slate Digital VCC N**e emu soundz real Gooooood!!

Postby ngarjuna » Wed Aug 25, 2010 8:24 pm

It seems hubristic to me to suggest that audible precision through modeled algorithms is impossible. It may be impossible with current processing/resources...that's not hard to believe. But sooner or later processing is not going to be much of an issue. At that point someone should be able to code something that, while maybe not identical in how it gets there, sounds pretty much the same, no?

My understanding was that where Nebula succeeded was its ability to capture these details in a way that's actually process-able with current technology, details that otherwise would be impossibly heavy to try to account for in algorithmic code. That shouldn't automatically mean that those details couldn't ever be coded. Or am I missing something?
User avatar
ngarjuna
Expert
Expert
 
Posts: 778
Joined: Tue Mar 30, 2010 5:04 pm
Location: Miami

Re: Slate Digital VCC N**e emu soundz real Gooooood!!

Postby farjedi » Wed Aug 25, 2010 9:10 pm

Can't you write code that mimics random behaviour?..it seems strange that you couldn't, I mean to the level that Nebula can..would love some more info..cheers
farjedi
User Level X
User Level X
 
Posts: 139
Joined: Thu Apr 01, 2010 7:12 pm

Re: Slate Digital VCC N**e emu soundz real Gooooood!!

Postby TranscendingMusic » Wed Aug 25, 2010 9:37 pm

Let me clear an aspect of this up here. This is less about code - because it is still all in the digital domain - and more about the implementation to traverse closer to realism. Where as algorithmic code is stricter definition an emulation, Nebula is more of a miner of characteristics of gear. It's not emulating, it's embodying.
.
When speaking with Giancarlo one time a while ago we were going over different descriptions about Nebula and one cool example he mentioned was something like this:
Let's suppose we have a photo of something, the photo featuring distortion for any given input level. So for example if audio is -35dB you apply one particular group of photo/snapshot/set of samples, if audio is -34dB you apply a different set of samples. The beauty of nebula is this: since it's applied to a group of samples and it's applied to blocks, its contingent on the delimiters of the blocks, so the switching point in nebula can't be forecasted. These blocks, as technology allows us, can contain so many variables being applied in a lateral sense rather than an already calculated code that goes from one prescripted stated, then to the next in a more linear fashion.
Now think of this, the stronger computing gets the closer and closer Nebula will get to the nature of analog. The code can only do what code can do to a certain extent. Please, though once again, the question of "if what alg.code is capable of doing at this point is enough or not" is another subject matter altogether.
.
Knowing what Nebula does already vs. what the newest coded plugin can do is the point here: that Nebula consists of a realm of realism that can not be paralleled - in a technical sense, not practically speaking. Just simply put, being truthful of what each one delivers; the different worlds they reside in. So saying this alg coded plugin is the first of its kind to do this or that - AS COMPARED TO OTHER ALG CODED PLUGINS, this is more or less okay. But to make the declarative statement that it is something new and unrivalled as an analog simulation compared to anything else - this is hyperbole at best IMHO.
mixing | mastering
Win 10 x64 | Sonar Platinum x64 | 3930K(OC)
User avatar
TranscendingMusic
Expert
Expert
 
Posts: 1132
Joined: Sat Mar 27, 2010 7:01 am
Location: USA

Re: Slate Digital VCC N**e emu soundz real Gooooood!!

Postby Mercado_Negro » Wed Aug 25, 2010 9:44 pm

I'm curious about something, how can people be so sure this isn't possible in DSP code (now or in the future)? Christians condemned Galileo and many others for their revolutionary ideas, hell some people even condemned Giancarlo in the past for Nebula (and some still do)! What would you do if you threw this VCC plug-in in one of your old mixes and it sounds better? Will you discard it because "it is not possible in DSP"? I'd think "hey this thing works in THIS mix, let's remix it" :)

Cheers
i7 3770k :: Asus P8H77-V LE :: 16Gb DDR3 @1600MHz :: Geforce GT 520 :: OCZ-Vertex 128Gb :: WD Black Series 1Tb and Green Series 1Tb :: F*******e Liquid56 :: REAPER 64bit and StudioOne 64bit (both latest versions) :: Win 10 64bit
User avatar
Mercado_Negro
Beta Tester
Beta Tester
 
Posts: 1353
Joined: Sat Mar 27, 2010 9:30 am

Re: Slate Digital VCC N**e emu soundz real Gooooood!!

Postby TranscendingMusic » Wed Aug 25, 2010 9:54 pm

Mercado_Negro wrote:I'm curious about something, how can people be so sure this isn't possible in DSP code (now or in the future)? Christians condemned Galileo and many others for their revolutionary ideas, hell some people even condemned Giancarlo in the past for Nebula (and some still do)! What would you do if you threw this VCC plug-in in one of your old mixes and it sounds better? Will you discard it because "it is not possible in DSP"? I'd think "hey this thing works in THIS mix, let's remix it" :)

Cheers


As far as I understand it, it's an actual limitation of traditional algcode. The way I like to think of it is, not thinking in terms of emulating a sound by a static state but reflecting a sound with its analog mechanistic realism intact; hearing an actual part of analog nature not copying it.
.
I'm not trying to speak for anybody else here either, I've already stated that practically speaking, there may be a certain level - regarding our perceptive thresholds - in which algcoded plugin suffices. May be, who knows?
My plight here is really a matter of what is being said and suggesting unprescedented tools where such tools have already existed (such as Nebula). I'm not putting out there which is better perceptively, subjectively or practically. I'm just saying that in examining and comparing on a technical level, one method of processing involves a mutli-variable function rather than a linear static state of functions.
mixing | mastering
Win 10 x64 | Sonar Platinum x64 | 3930K(OC)
User avatar
TranscendingMusic
Expert
Expert
 
Posts: 1132
Joined: Sat Mar 27, 2010 7:01 am
Location: USA

Re: Slate Digital VCC N**e emu soundz real Gooooood!!

Postby ngarjuna » Wed Aug 25, 2010 10:01 pm

TranscendingMusic wrote:When speaking with Giancarlo one time a while ago we were going over different descriptions about Nebula and one cool example he mentioned was something like this:
Let's suppose we have a photo of something, the photo featuring distortion for any given input level. So for example if audio is -35dB you apply one particular group of photo/snapshot/set of samples, if audio is -34dB you apply a different set of samples. The beauty of nebula is this: since it's applied to a group of samples and it's applied to blocks, its contingent on the delimiters of the blocks, so the switching point in nebula can't be forecasted. These blocks, as technology allows us, can contain so many variables being applied in a lateral sense rather than an already calculated code that goes from one prescripted stated, then to the next in a more linear fashion.


But something algorithmic should be able to do this too. The feature which makes it currently "impossible" is that you would need effectively different code for all the different possible programatic inputs. We're in agreement that, right now, this would make the code pretty much unusable and unwieldy. And that is, likewise, the genius of using volterra series to keep track of all those different possible inputs: it's able to model all those different levels of non-linear behavior without being so bogged down as having to reinvent the wheel for each different possible input parameter.

However...on a purely theoretical level...even Nebula has limitations here, to the number of kernels I suppose (admittedly this part of the conversation is getting beyond my expertise). Can Nebula really discriminate between -10db and -10.00001dB input in terms of the non-linear differences between the two? Or is it just that Nebula is stepped enough that the major (as in audible) potential inputs all use different samples to calculate the processing? I truly don't know the answer to that question. My understanding (and I'm happy to be corrected if this is wrong) is that Nebula is "99%" there because it's able to apply enough variance based on input, but it's still not 100% non-linear the way analog input would be.

It should, then, be possible when processing resources allow to do just that with algorithms, having different processing based on a series of non-linear differences dependent on program input. It would be an ugly mess I'm sure compared to using volterra series...it might be impossible to find a computer beefy enough to actually run such code for a long while...but it shouldn't be theoretically impossible by any means.

Knowing what Nebula does already vs. what the newest coded plugin can do is the point here: that Nebula consists of a realm of realism that can not be paralleled - in a technical sense, not practically speaking.


I would only suggest that Nebula consists of a realm of realism that has never yet been paralleled. I just don't understand why it's theoretically impossible for something to come along that would parallel it. I'm not suggesting Slate consoles are that something, I actually suspect they are not; but I think we are potentially underestimating the ingenuity of other developers if we think it will never happen.

Maybe it's not worth reinventing that wheel; maybe using volterra series math (which is especially well suited to this particular task, which I'm sure was inspirational to Giancarlo et. al in the development of Nebula in the first place) is so much easier that it would just be stupid to try and create algorithms that could account for the same kinds of variables. But that's still not the same thing as saying it's technically impossible.

Just simply put, being truthful of what each one delivers; the different worlds they reside in. So saying this alg coded plugin is the first of its kind to do this or that - AS COMPARED TO OTHER ALG CODED PLUGINS, this is more or less okay. But to make the declarative statement that it is something new and unrivalled as an analog simulation compared to anything else - this is hyperbole at best IMHO.


No question, you have to separate the marketing hyperbole from the real capabilities of a tool. Again, I'm not necessarily interested in vindicating any current product, I'm simply questioning the notion that this can never be done with other coding methods than volterra series.
User avatar
ngarjuna
Expert
Expert
 
Posts: 778
Joined: Tue Mar 30, 2010 5:04 pm
Location: Miami

Re: Slate Digital VCC N**e emu soundz real Gooooood!!

Postby giancarlo » Wed Aug 25, 2010 10:06 pm

HIFIDELITY wrote: Is giancarlo right to not trust demos because who knows how the demos were created


really I didn't want to accuse them or nobody else, especially in THIS specific case. I was speaking about audio demos in general. During 2006 I had a discussion with christian budde on kvr and I had the temptation to post directly an analog filter instead of the new mistery plug-in, so I know perfectly how a developer feels when he is attacked in some way. For what I know, someone is still tweaking demos, and I would not be surprised very much about that: even top models are "fixed" with a bit of photoshop in most of cases.

About the random behaviour, we could open a topic regarding it. Nebula is deterministic, not random. But the result is random, due to the amount of variables and the behaviour of the vectorial engine: its period doesn't start always at the same point, even if you render the same audio twice. So it's a consequence. Mimic a random behaviour is easy, though.

Again: you could emulate harmonics preatty easily (and you couldn't distinguish them using a bode diagram), what's difficoult is to get exactly the same time behaviour. Here we are speaking about details, if you are not interested in details you could even disable harmonics (or reduce the kernels size) and you could run A LOT of alexb instances. I think NOBODY will ever detect it inside a busy rock mix. We could post a couple of demos and get out a quick acqua plugin, and tell everybody we nailed it.

It took around 10 years to see posts on public forums regarding nyquist issues with iir equalisers. This was a well known issue among the developer community. Now you know that a lot of plugins are not behaving correctly there, and that you need a fir implementation in most of cases, otherwise you need to be CAREFUL there, a stock biquad algorithm does NOT sound as much good as expected to be. Well, nobody was even speaking about that, users thought first eq emulations were a perfect replacement of hardware. They were more interested in noise and other silly things at that time. After some time, SOMEONE posted that his vintage eq WAS different and high boosts were better.

You could imagine what is missing in current dynamic harmonic distortion implementations, where nobody has a clear knowledge (you should do volterra kernels analysis for example, but it seems like knowledge is not very good on this matter, because teams ARE small, this is still a niche market).

so... everything is a matter of attitude, nothing less, nothing more. Here we can't speak about the product because it's too early. And if you don't believe that, sample a real M****y and post a couple of demo files on gearslutz (or render a couple of files in nebula if you are lazy), post a picture of a possible gui (maybe rendered in 3d max studio), and tell everybody that developer xxx is working for you and he nailed it perfectly, and that hardware soon will be a piece of unuseful iron, than just take some pop corn.
User avatar
giancarlo
Founder
Founder
 
Posts: 9198
Joined: Mon Sep 21, 2009 10:40 pm
Location: Italy

Re: Slate Digital VCC N**e emu soundz real Gooooood!!

Postby TranscendingMusic » Wed Aug 25, 2010 10:12 pm

Yes whether or not something could come along to parallel Nebula is another question altogether and that of course I don't think can be answered definitively at this juncture.
At the risk of sounding like a broken record and coming across as annoying, and I apologize if I have already, :? it would be great Giancarlo if you could expand on these fundamentals even more to help clear the concept of algcode vs. Volterra Kernels.
ADDED: But so far simply put, algcode: one state of function that determines the next. Volterra: random behavioral results that can harbor the vast complexity of analog nature.
.
Regarding the Kernels: that's really just a matter of processing power too. Any variable and distortion Kernels as well can be tact on when it comes to nebula. It's just the technological limitations at this point that limit it.
mixing | mastering
Win 10 x64 | Sonar Platinum x64 | 3930K(OC)
User avatar
TranscendingMusic
Expert
Expert
 
Posts: 1132
Joined: Sat Mar 27, 2010 7:01 am
Location: USA

Re: Slate Digital VCC N**e emu soundz real Gooooood!!

Postby ngarjuna » Wed Aug 25, 2010 10:26 pm

TranscendingMusic wrote:Yes whether or not something could come along to parallel Nebula is another question altogether and that of course I don't think can be answered definitively at this juncture.
At the risk of sounding like a broken record and coming across as annoying, and I apologize if I have already, :? it would be great Giancarlo if you could expand on these fundamentals even more to help clear the concept of algcode vs. Volterra Kernels.
.
Regarding the Kernels: that's really just a matter of processing power too. Any variable and distortion Kernels as well can be tact on when it comes to nebula. It's just the technological limitations at this point that limit it.


Absolutely. And I wasn't meaning to be pedantic and suggest "Someday, something even more incredible will come out!!!" because, surely, we all know that. I think Giancarlo explained what we've been dancing around when he said:

...About the random behaviour, we could open a topic regarding it. Nebula is deterministic, not random. But the result is random, due to the amount of variables and the behaviour of the vectorial engine...


That tells me that you could indeed build an engine which would be able to do what Nebula is doing using something other than volterra series. However using dsp (as we currently understand it) would require SO MUCH code that it would be practically unusable. The volterra series allows you to make fast use of a large number of variables that current processing power would not allow using dsp style code. I think this is the beauty of the volterra series in modelling non-linear behavior; it's not the only way you can do it, but it's so efficient that you can actually apply it practically (real time) to audio.

Unless I'm misunderstanding Giancarlo, in which case hopefully he'll correct me.

I think you're right, by the way, about how future processing milestones will affect Nebula. What's already 99% will continue to get closer and closer to the limit.
User avatar
ngarjuna
Expert
Expert
 
Posts: 778
Joined: Tue Mar 30, 2010 5:04 pm
Location: Miami

Re: Slate Digital VCC N**e emu soundz real Gooooood!!

Postby TranscendingMusic » Wed Aug 25, 2010 10:30 pm

ngarjuna wrote:
TranscendingMusic wrote:Yes whether or not something could come along to parallel Nebula is another question altogether and that of course I don't think can be answered definitively at this juncture.
At the risk of sounding like a broken record and coming across as annoying, and I apologize if I have already, :? it would be great Giancarlo if you could expand on these fundamentals even more to help clear the concept of algcode vs. Volterra Kernels.
.
Regarding the Kernels: that's really just a matter of processing power too. Any variable and distortion Kernels as well can be tact on when it comes to nebula. It's just the technological limitations at this point that limit it.


Absolutely. And I wasn't meaning to be pedantic and suggest "Someday, something even more incredible will come out!!!" because, surely, we all know that. I think Giancarlo explained what we've been dancing around when he said:

...About the random behaviour, we could open a topic regarding it. Nebula is deterministic, not random. But the result is random, due to the amount of variables and the behaviour of the vectorial engine...


That tells me that you could indeed build an engine which would be able to do what Nebula is doing using something other than volterra series. However using dsp (as we currently understand it) would require SO MUCH code that it would be practically unusable. The volterra series allows you to make fast use of a large number of variables that current processing power would not allow using dsp style code. I think this is the beauty of the volterra series in modelling non-linear behavior; it's not the only way you can do it, but it's so efficient that you can actually apply it practically (real time) to audio.

Unless I'm misunderstanding Giancarlo, in which case hopefully he'll correct me.

I think you're right, by the way, about how future processing milestones will affect Nebula. What's already 99% will continue to get closer and closer to the limit.


Yea definitely! Thanks G for chiming in... And the "determinstic behavior" was definitely meant as the result not how it functions. Of course anything coded, in the digital domain, has a determined behavior :)
Yea and the limit when it comes to perception, could be lower than we even imagined. Our perception is the "speed limit" so to speak.
mixing | mastering
Win 10 x64 | Sonar Platinum x64 | 3930K(OC)
User avatar
TranscendingMusic
Expert
Expert
 
Posts: 1132
Joined: Sat Mar 27, 2010 7:01 am
Location: USA

PreviousNext

Return to Other things

Who is online

Users browsing this forum: No registered users and 1 guest