Aquamarine

Share your experience, tips & tricks, show off your music.
User avatar
Dimzo
Member
Member
Posts: 201
Joined: Tue Oct 23, 2012 4:19 pm

Re: Aquamarine

Post by Dimzo » Sat Jan 17, 2015 1:29 pm

brilliant news!
time for a new demo vid :mrgreen:

just kidding, take your time to finish this thing so we can tweak it ourselves ;)

Mplay
Member
Member
Posts: 294
Joined: Fri Aug 20, 2010 10:25 am
Location: Willemstad, Curaçao

Re: Aquamarine

Post by Mplay » Tue Jan 20, 2015 6:45 pm

Offly quiet in here. Silence before the storm?

SWAN
Expert
Expert
Posts: 775
Joined: Fri Mar 26, 2010 11:16 pm

Re: Aquamarine

Post by SWAN » Tue Jan 20, 2015 7:42 pm

Giancarlo is working on the beta so no demos yet....beta testers waiting on an update...
Mac Mini i7 quad 2.6

Logic X
Live 9
Reaper

User avatar
giancarlo
Founder
Founder
Posts: 11215
Joined: Mon Sep 21, 2009 10:40 pm
Location: Italy

Re: Aquamarine

Post by giancarlo » Tue Jan 20, 2015 8:06 pm

we are working hard guys.
Most of time bugfixing. I want to check that all shapes are properly matched.
Before the beginning of the previous week I was focused on adding features, keeping the same compression action we got in december (created using the same technique used in titanium release)
So we improved meters (still working on the gui atm), we added external sidechain, fixed several crash bugs and so on.

Then I worked on the tech, improving envelope followers (now they are smoothed).

Yesterday I started debugging shapes, I'm working on bugs more or less. For example I discovered that shapes were resetting their internal time, so the correct shape was not generated properly (because they are based on sampling).
Till I'm not sure about them (I need to bug fix the engine properly) I will not release to betatesters this new version.
Even tested on audio apparently everything is fine, but till I find bugs I know it is not ;)

Brian
Vip Member
Vip Member
Posts: 372
Joined: Fri Sep 24, 2010 5:10 pm
Location: Taipei, Taiwan / San Francisco, CA

Re: Aquamarine

Post by Brian » Wed Jan 21, 2015 2:28 pm

Keep up your efforts to push the technology forward!

I sometimes mix and record in a studio with the hardware and find it very versatile as a submix and track compressor too. Even distorted guitars or saturated, layered synths in harder EDM styles (despite having little dynamic range) can benefit from the tone. The optical compressor with steel transformer and makeup gain adds some awesome color to aggressive sounds like this. I've just pre-ordered so I can use multiple instances in a mix, and not have to bounce, etc.

One question - how is the discrete/vca makeup gain working? Will increasing it up also drive the output/transformer section harder? Has this gain stage been sampled or is makeup achieved with clean digital gain?
Official dealer of Acustica Audio products in Taiwan.
Acustica Audio台灣授權經銷商
[email protected]

Mixing/Mastering Engineer at Sentient Sound Labs 覺知音樂工作室

Win10 x64 | Reaper5 | PT12 | Ableton Live 9

User avatar
giancarlo
Founder
Founder
Posts: 11215
Joined: Mon Sep 21, 2009 10:40 pm
Location: Italy

Re: Aquamarine

Post by giancarlo » Wed Jan 21, 2015 8:23 pm

ok, I'm working on aquamarine exactly now! and I'm debugging the new tech developed for aquamarine

First of all, answers to your questions, and I hope a lot of answers to all people who already bought it (many!!!! thank you for the support!)
how is the discrete/vca makeup gain working?
first of all this compressor works at -18dBFS. It means that you have standard operation at -18dBFS, and you have plenty of room for pushing it really hard.
The normal unit compresses a lot indeed if you push it really hard with hot inputs.
In our assessments (we have kept all audio files) the discrete compressor reached 65dB+ of compression. So my main issue was to keep this behaviour for crazy usage.

Later we discovered that even our expert betatesters were a bit confused by the -18dBFS thingy. We are not speaking about a RMS value, but a peak one, so really the input was too much quiete for a normal usage.

So we implemented a trim input knob instead of a simple input one, the "input trim" was suggested by our beta team.

It means you have a knob which by default is turned to -18dBFS. So you can use this compressor like a normal software compressor like titanium, it works at 0dBFS (maximum peak audio level).

Than if you want to DRIVE it you just turn the knob clockwise.
Levels will not change, but you will simulate an hotter input.
Will increasing it up also drive the output/transformer section harder?
Here a bit of tech. The preamp in aquamarine is positioned at the beginning of the chain instead at the end, but it is NOT a mistake. It is maths.
Infact we sample a chain as a whole, not only the preamp but also the compressor tone.
If you split the effect in tone + gain computer it is important the tone is before the compressor, so it changes accordingly with input level (before the gain modification) and it takes in account the post-compression gain, saturation and effect.
Just to use different words, for us the distortion chain was sampled as a whole, indipendently from the obvious compression effect.

Let's make an example: you drive the compressor a lot. The compressor would drive accordingly the preamp in a real hardware compressor. But we sampled as a whole! so we have the preamp effect based on the input level and other parameters like ratio, which makes the overall preamp tone, so the position in our case is correct at the beginning (because we didn't sample the only preamp but the whole chain).
Has this gain stage been sampled or is makeup achieved with clean digital gain?
everything was sampled this time. Tone of our preamp is the usual tone with nebula quality, it means harmonic distortion achieved by fir filters, so the exact match of magnitudo and phase also for harmonics. In this field nebula is quite unique, and it is the reason of its high-end sound.

As for titanium, (but even better) we worked on our tech in order to provide correct shapes, which is not very common in the digital world. If you analyze most of digital compressors you will discover that shapes are "cook-book" ones in most of cases.

For example they cannot match easily a shape like this one
(DISCLAIMER, BMANIC WILL KILL ME BECAUSE HE HATES WHEN I POST WAVELAB PICTURES, I ASK MERCY)
REL 0.png
REL 0.png (6.75 KiB) Viewed 3265 times
now if you ask to me if they null I'll tell you immediatly, we are going to reduce the error at each release, but they honestly don't null. It is something very difficoult to null in the software domain because it is a function of an awful lot of variables.
But hey, we are DAMN close. It is not simply a matter of TIMES. It is a matter of SHAPES. Most of companies are just trying to emulate timings. Which are pretty enough for a proper sound, but not for me/us. So we are going to reach this precision at an acceptable aliasing, in realtime, and smoothing
We implemented a new generation of envelope followers in nebula and they are based on a proper smoothing. And things will be improved further in the future, because now we are limited to processing audio each 2 samples, we are going to remove this limitation. Anyway, even with this limitation result speaks by itself
aliasing.png
How to read it: harmonics are there as they should be. There is no aliasing (no strange things on the left of the first tone). The only imperfection is the sidelobe on each harmonic (the cone), but it is at a safe level (just to make it clear, when something is located below 100dB we are in a safe region). When the engine will run exactly at samplerate it will disappear, it is my obsession now :D

I think it is pretty good for an hardware emulation which is trying to mimic the envelope shape by sampling. The excercise would be simpler using a cookbook shape and a digital/clean tone.

And now I have a surprise.
You know that nebula is able to create program-dependant compressors based levels. Tim p showed it in his 1176 emulation. Titanium is not program dependant at all, since the hardware is not.
We created a program-dependant emulation based on target level.

But we thought it was not enough, since the original beast is program dependant on the time too.

So this time we decided to improve things to a crazy level, and this is the result:
PROGRAM DEPENDANT.png
As you can see times change accordingly with the size of the input charge time.

Not only; if you are asking about the different smoothing, it depdends on the ratio level: ie the same pictures at an higher ratio:
linearity.png
linearity.png (11.38 KiB) Viewed 3265 times
last one is quite "linear" shaped
last.png
last.png (3.06 KiB) Viewed 3265 times
as you can see, a very very very complex model with a really particular sound and a lot of details.

Uh, very last note. Do you know what happens on hardware when you drive it hot using a long testtone?
this thingy:
very last.png
very last.png (5.72 KiB) Viewed 3265 times

lol

Brian
Vip Member
Vip Member
Posts: 372
Joined: Fri Sep 24, 2010 5:10 pm
Location: Taipei, Taiwan / San Francisco, CA

Re: Aquamarine

Post by Brian » Wed Jan 21, 2015 8:58 pm

Thanks for the update and detailed reply! Yes, I understand now about the signal flow and your reasons for including the amplifier/transformer sampling as the first stage (after input trim).

It seems this will be an important release for your Acqua line, both in terms of R&D but also hopefully offering an amazing sounding tool to get people more interested in the Nebula scene.

This should also be the most versatile Nebula compressor so far with four possible modes: opto, vca, combo, or simply amp coloration.

Very much looking forward to the release.
Official dealer of Acustica Audio products in Taiwan.
Acustica Audio台灣授權經銷商
[email protected]

Mixing/Mastering Engineer at Sentient Sound Labs 覺知音樂工作室

Win10 x64 | Reaper5 | PT12 | Ableton Live 9

RJHollins
Expert
Expert
Posts: 4047
Joined: Sun Mar 28, 2010 5:53 pm

Re: Aquamarine

Post by RJHollins » Wed Jan 21, 2015 10:25 pm

Thanks for taking the time for the details in the design/implementation.

I hope that a special manual displaying some of this info is included with the package. This type of info may not interest everyone ... but the tech engineers are always studying/learning ... particularly how the unit functions.

Great stuff ... 8-)
i7-5820k, MSI X99A Plus, 16 GIG Ram, Noctua NH-D14, Win-7 Pro [64-bit], Reaper-64

NVC [Nebula Virtual Controllers]

User avatar
Tim Petherick
Expert
Expert
Posts: 1774
Joined: Sat Apr 17, 2010 4:07 pm
Location: Bath , Uk

Re: Aquamarine

Post by Tim Petherick » Wed Jan 21, 2015 11:14 pm

This looks great!,

Since you mentioned dynamic dependence on u76 Giancarlo hope you don't mind me adding, it was my intention to create time dependence on u76, this attempt is on the special AT presets. I created this in a non official way of course.....I created it through release time being controlled through a extra EVF with release times simulating a charge.the graphs on my page pretty much display this on my website. I'm
Not trying to say it's a super accurate way of doing it but it gives a disirable affect.

Of course I couldn't of done this with out rewiring a few elements. It was not based on previous templates.

By the way the release shapes look very good on the product using this new tech of yours.

bmanic
Expert
Expert
Posts: 865
Joined: Thu Apr 22, 2010 8:42 pm

Re: Aquamarine

Post by bmanic » Wed Jan 21, 2015 11:27 pm

giancarlo wrote: As for titanium, (but even better) we worked on our tech in order to provide correct shapes, which is not very common in the digital world. If you analyze most of digital compressors you will discover that shapes are "cook-book" ones in most of cases.

For example they cannot match easily a shape like this one
(DISCLAIMER, BMANIC WILL KILL ME BECAUSE HE HATES WHEN I POST WAVELAB PICTURES, I ASK MERCY)
REL 0.png
I won't kill you for the picture (though I do hate Wavelab and it's incredibly crappy aliased graphics!) but I will remind you of toning down the arrogance. It is NOT true what you said about compressor shapes. It's pretty much exactly the opposite with today's complex analogue emulated plugins. The actual compressor shapes are extremely close to the real hardware!!!. I really would like it if you bought a HUGE mirror and looked into it and said: "I will stop being arrogant and not speak silly about other developers and make up "facts" that aren't at all true".

Please. Do that.

If there's one thing that many other companies do correctly, it's the actual shape of the compression.. both attack and release. IK Multimedia 1176 is absolutely excellent. Heck they even got the "all buttons down" working very close to how our hardware works.. not exactly the same but close enough for most people to not hear any real difference.

Another good example is Cytomic's The Glue. It is very very very very close to the real thing when it comes to the shape.. it just doesn't model anything else than the compression. Slate Digital's 1176 is pretty damn close to his own 1176 that they modeled. UAD's latest 1176 and Fairchild and LA2A are ridiculously closely matched to the real hardware when it comes to the shapes and general behavior.

What they are NOT getting correct, and where Nebula/Aqua is king, is the "tone" and overall "realness" of the global sound of a hardware unit. It probably comes down to the phase and complex harmonic structures which are really hard to get algorithmically correct.

Also the Elysia Alpha Compressor and Mpressor are incredibly close to the real hardware in terms of attack and release times.

So yeah, in short: You are completely and utterly wrong.. about 10 years behind development. What you just said about "cookbook shapes" was perhaps true 10 years ago but it's not at all true any more.

bmanic
Expert
Expert
Posts: 865
Joined: Thu Apr 22, 2010 8:42 pm

Re: Aquamarine

Post by bmanic » Wed Jan 21, 2015 11:36 pm

As for complex attack and release shapes.. You want complex?

I'm under NDA so I can't disclose any details but lets just say that my algorithms in FabFilter Pro-L have a ridiculous amount of elements, each completely interactive, that constantly vary every single element of internal components.. attack timings, release timings, various "thresholds" and knee curves and very complex and constant shape modulation of pretty much everything.. so yeah. No cook book there.. it's hundreds of times more complex internally than any analogue hardware could even dream of ever being.. and we got it all crammed into a few control knobs. :lol:

User avatar
kj.metissage
User Level V
User Level V
Posts: 50
Joined: Wed Mar 19, 2014 5:42 pm
Contact:

Re: Aquamarine

Post by kj.metissage » Wed Jan 21, 2015 11:43 pm

bmanic wrote:As for complex attack and release shapes.. You want complex?

I'm under NDA so I can't disclose any details but lets just say that my algorithms in FabFilter Pro-L have a ridiculous amount of elements, each completely interactive, that constantly vary every single element of internal components.. attack timings, release timings, various "thresholds" and knee curves and very complex and constant shape modulation of pretty much everything.. so yeah. No cook book there.. it's hundreds of times more complex internally than any analogue hardware could even dream of ever being.. and we got it all crammed into a few control knobs. :lol:
Sounds like "Pro-L 2" is in development. Yay !!!

bmanic
Expert
Expert
Posts: 865
Joined: Thu Apr 22, 2010 8:42 pm

Re: Aquamarine

Post by bmanic » Wed Jan 21, 2015 11:53 pm

It is? They didn't tell me anything!!

No seriously, I don't work for FabFilter. Pro-L was just a one time thing. I do beta test for them though but I don't know of any future products any more than you guys.

Latest product was Pro-Q 2 and I think they are now on a well deserved vacation.. or plugging along in their dungeon creating new awesome products.

But hey, we are here for Nebula/Aqua plugins. A completely different type of thing (I'd call it utterly revolutionary even!!).

I just wanted to illustrate the thing about attack and release shapes. Latest plugins from UA, Slate, IK Multimedia, Brainworks, Softube heck probably even Waves.. are complex things. They are not "simplified" or "cook book" anything. There's real science and truly hard work behind them all and they deserve some respect for that in my opinion.

User avatar
kj.metissage
User Level V
User Level V
Posts: 50
Joined: Wed Mar 19, 2014 5:42 pm
Contact:

Re: Aquamarine

Post by kj.metissage » Thu Jan 22, 2015 12:05 am

Alright man ! You're right.

The latest waves (no pun intended) of native plugins have been incredible.

As are the latest Acqua releases.

User avatar
giancarlo
Founder
Founder
Posts: 11215
Joined: Mon Sep 21, 2009 10:40 pm
Location: Italy

Re: Aquamarine

Post by giancarlo » Thu Jan 22, 2015 12:06 am

about cookbook It was not addressed to a specific developer, ie I did not test uad and vmr yet.
It happens I see very often wrong shapes, though, more often than expected
In a lot of cases they do not change, for example - they are just sort of stretched

Post Reply