Three-Body Technology Cenozoix Compressor (Plugin Alliance)

[Ploki]

Thank you! I’ll give it a go after i finish work

[MirkoVanHauten]

@Ploki

Here is a very quick and dirty example of some compression that I find very difficult to do with other tools. Notably listen to just how much and how "cunky" (7:1 ratio mostly) the compression is, while all the transients still sound really smooth and rounded.

Original Audio Source
Cenozoix doing Chunky/Punchy compression
Cenozoix doing pumpy/strained compression

(All links above are Google Drive links, 48kHz 24bit.. that I possibly forgot to dither. )

I'm a bit tired after a long day of work but hopefully this is a good starting point in demonstrating in what I mean by it being able to handle difficult sources without introducing a lot of clickyness/"digital edges" while still being really upfront and punchy.

What were the settings you've used? I just wanted to quickly compare it other comps to hear that "click".

[Ploki]

@Ploki

Here is a very quick and dirty example of some compression that I find very difficult to do with other tools. Notably listen to just how much and how "cunky" (7:1 ratio mostly) the compression is, while all the transients still sound really smooth and rounded.

Original Audio Source
Cenozoix doing Chunky/Punchy compression
Cenozoix doing pumpy/strained compression

(All links above are Google Drive links, 48kHz 24bit.. that I possibly forgot to dither. )

I'm a bit tired after a long day of work but hopefully this is a good starting point in demonstrating in what I mean by it being able to handle difficult sources without introducing a lot of clickyness/"digital edges" while still being really upfront and punchy.

these sound good
But i need an example with clickiness! i'm still not sure what you mean with that.

Could you share the cenozoix presets for these two? I wanna play around a bit

thanks for your time

[DCrown]

Can't you go for a walk somewhere in the rain-forrest or something?

Is stalking your only hobby?
Or does contributing to a thread mean
hyping a product without comparing it for you and hiding the downsides ?!
Maybe you are just not clever enough to see there could be some truth in irony...
Again a great contribution from your side to a topic, as expected.
I also feel for you, it obviously hurts you if someone doesn't like something you like, now imagine how hard my life is, there's so much hate for Waves and there are people who don't like Avenger, my two fav devs/plugins.

[audiogeri]

Can someone please confirm that it is too CPU hungry for using a lot of instances on individual tracks?

[ToMegaTherion]

[quote=bmanic post_id=8844715 time=1707330323 user_id=5744]
[quote=Ploki post_id=8844636 time=1707325426 user_id=221865]
Declick is cool, but Pro-C2 doesn't click. When it does, 2ms lookahead solves it

[/quote]

Nope. I don't know what you listen to but I'm not talking about an overshoot click. It's different. I'm talking about the typical "plugin compressor" click, something that just isn't present in most hardware. It's one of those areas where hardware just does way better.

This is the first digital compressors that can do the smash but not be "clicky" in a digital way. There are of course smooth compressors like Unisum etc that can be super fast and still not make any clicks but the downside is that when you set them up in that way, they become smooth and not at all chunky.

This is the only compressor where I've been able to do super chunky (think of it as "macro clicks") while still being completely micro-click free.

.. hard to explain but I'll try to post some audio examples to show what I mean.

Especially fast smashy compression (think DBX160 style) where you really make things "plop!" hard, is where most digital compressors fail. There's always this added clickyness to each transient that is absent in hardware. It may be a very short sudden burst of aliasing or intermodulation that just isn't smooth.. but that's just speculation on my part. Anyhow, you CAN get rid of that clickyness in some plugins like DMG Audio Trackcomp 2 but it requires huge amounts of oversampling to do so. With Cenozoix you can get rid of it with a single knob and no oversampling at all. It's absolutely fantastic! A bit like having TDR de-edger plugin built into it, except to my ears Cenozoix method sounds even better than de-edger.
[/quote]
The digital compressor clicks are from the peak detector not being being able to detect the peak (and thus the rms if rms detectors) of the signal in single sample rates. They have to emulate, well try to perform, the transform of the da converter to obtain the true peaks. 2-4x upsampling alone is not enough for many signals. See ITU BS.1770 paper with its estimated overshoot at various sample rates, which seems to be still off given the ice picks I have heard from every 1176 emulation. Thus without true peak detection, they cannot determine the polarity of the signal and thus switch between attack and release. Distortion in the side chain will also make the multiplier do stupid things to the audio.

The TDR compressors do not click on insane at faster attack. Klanghelm dc8c3 at 8x oversampling doesn’t either if set to smash and enough feedback. PSP Impressor didn’t either with the 16x oversampled sidechain.

Some of the processors with short lookaheads do not click too. The Softube Weiss DS-1 only clicks when you set attack and release so fast as to send it into oscillation or cause weird discontinuities in the switching between them. Oxford Dynamics (20 samples) and Pro C2 with a fraction of a millisecond of lookahead don’t either and can go pretty fast but they reach a point where they cannot speed up to my ears, Oxford Dynamics on release and Pro C2 on attack. The Oxford Dynamics can get the smoothness you mentioned along with Pro Audio DSP DSM, MDWDRC2, and Sound Radix Powair. The issue is lookahead slows down the rate of maximum modulation of the compressor by ramping the attack (or having a hold), prevents the compressor from compressing high frequencies because it modulates slower than their cycles unless the lookahead is very short like the Weiss DS-1’s default of 0.02 ms.

Analog can click too btw. The first transient to cross threshold of a dug in too much feedback compressor will often click horribly because its reacting to its own processing of the signal rather than the signal itself. A deeply dug in 1176 will misfire. Feedforward compressors can often leave plastic transients from lack of program dependency in poorer designed compressors but at least they react to the signal. DBX 160 series are some of the worst offenders because they expand when crossing threshold and then clamp down based on the sort of RMS with a slow attack and fast release unless you overcompress. They usually need to be limited afterward ime but at least RMS detectors are lower distortion than peak detectors. The same with the peak compressors with an attack curve that lets most of the initial peak pass through without even being shaped.

[woodsdenis]

Looks like what fabfilter pro-c3 couldve looked like lol

I wrote this before i realised it's the same company behind Kirchoff EQ.


Ok so three body tech appears to be on a FabFilter mission.

They take a proven design and layout from FabFilter and cram a fuckload of extra functionality in.

So naturally, i'm exploring this thing mostly as a Pro-C2 alternative first, because the control ranges, layout, GUI, M/S modes and operation/linking, sidechain layout and functionality are all very obviously "inspired" (putting it lightly) by Pro-C2.


From the get go, this thing uses more CPU than Pro C2, when both on clean mode with no oversampling.
With no oversampling, Cenozoix is a smidge cleaner.
With oversampling @2x they're similarly clean. However, Cenozoix uses so much more CPU than ProC2 when oversampled it's not even a comparison.
@2x oversampling, Pro-C2 uses about 7% of my CPU, 10% with UI open, and Cenozoix uses = 50%, 70% with GUI open, nearly 10x MORE.
(I have istat menus set so 1 core is 100%)
And not on the account of being cleaner! Pro-C2 is still cleaner and produces less THD with similar GR and settings...

The Analog models mean nothing to me without suggested ranges like TrackComp has them. (I don't use track comp btw), there's a fundamental principle of interaction with topology that makes a specific sound. It's hard to make Cenozoix sound like a VariMu, because the controls make no sense for someone used to how a VariMu behaves.
I can get Virtual Mu mode Cenozoix to sound close to Pulsar Mu, but the controls to get there make absolutely no sense.
I don't use Mu that way. I use it to quickly get the VariMu sound with the controls that make sense for VariMu, not endless hunt too many parameters so i can sort of make it respond like a Mu.
TrackComp has this nailed down with original ranges with ability to go extended.

FET modes are also decent, but they're completely unintuitive to work with if you worked with any 1176 in your life.

I've played around with it, thrown a few vocals, mixes and some test tones in it and compared similar algos (mAstering ProC2 vs Mastering Mode Cenozoix) and for some amount of GR, I preferred Pro-C2 in most cases, as for analog models it's really hard to compare them. I did manage to smash drums nicely with Blue FET.

I do like the Punch/Pump slider. That thing is great.
There's also neat controls like Clamp.
But somehow with many less controls than Unisum, it's less intuitive than unisum.


GUI quirks:
for some reason release "Hold" is in the bottom bar. I don't get that, it's part of the VCA envelope.
Many menus and windows reopen when you reclick. i.e. you open presets and want to close them - you cant click again on the same preset button because it will REOPEN them, you need to click somewhere else. Or not. Sometimes they close. Really inconsistent.


FabFilter is designed through and through with every detail in mind and it's refined to perfection when it comes to user interaction. Thins thing has little finesse and feels thrown together.
In my opinion, it has the exact same problems as Kirchoff. Boasting some super features (Kirchoff boasted 128bit internal res) and tons of functions, but at the end of the day being a weird clone of Fabfilter.
You can't dethrone fabfilter by being a clone of fabfilter.

Pro-C2 from wish. I'll wait for Pro-C3.

Edit: btw i really wanted to like it because of @bmanic post
It doesn’t sound bad at all, neither did kirchoff.
It’s just not well thought out and with high cpu usage i really dont see what it’s purpose is. For track use it’s too CPU hungry and for mastering there’s better options imo.

No mad CPU issues here on a silicon Mac, haven't done a full scientific breakdown but it seems fine with a number of them on a session.

[Ploki]

Someone over at gearspace theorised that Declick is actually just a soft clipper applied to detected transients.

After trying out the concept by following turning the declick to 0% and following Cenozoix with various clippers, i can say it sort of makes sense. It did make the most sense when i followed it with Saturn set to push drive with the transient envelope follower.

.A simple ‘post-compression clipper’ can be used to reduce clicks, but Cenozoix’s adjustable De-Click parameter goes further. It detects the transients in real time to find the clicks, and sends them all to the specially designed overdrive effect, which transforms the amount of clicking transients into the amount of overdrive. As a result, the clicking sounds can be significantly reduced while keeping the power of the transients.

So declick appears to be a transient triggered clipper, and "clamp" appears to be a simple post-limiter.

Push/Pump however is pretty unique.
The only ITB compressor that i saw do the "pump" attack phase as aggressively is the Fuse Audio VCL-515.
The only other compressor i've seen having that attack "dip" is Molot in Sigma, but it's much less pronounced

Screenshot 2024-02-08 at 12.40.03.png

No mad CPU issues here on a silicon Mac, haven't done a full scientific breakdown but it seems fine with a number of them on a session.

16" M1 Pro/32gb

I didn't say i have CPU issues, i said it uses way more CPU than my usual go-to - especially when oversampled.

Single (mono) channel, non-oversampled, clean, CPU usage is for both Logic idle + Tracks with compressors (logic takes about 20% of a core idling + another few for noise generation)

Cenozoix runs 300 instances at 900%
Pro-C2 runs 300 instances at 110%


For Pro-C2, i can run 1080 instances (similar settings) and it hits about 360%. I could keep adding more but i got bored, point proven 500 instances ago.

That's such a big difference in optimisation it's not even a contest.
Also Pro-C2 loads up much MUCH faster. If I load 10 Pro-C2 at once it takes a fifth of a time to load it compared to cenozoix.

Now don't barge in with "BuT WhO NeEdS 300 cOmPreSsoRs" - nobody does, but I like to know which of my tools are capable of careless throwing shit around and which will become a problem in larger sessions. I know i can rely on Pro-C2 to not even make a dent into CPU, i can track with it at no latency, and i know it won't bog down large sessions if i have large quantity of them.

[jtsterays]

Push/Pump however is pretty unique.
The only ITB compressor that i saw do the "pump" attack phase as aggressively is the Fuse Audio VCL-515.
The only other compressor i've seen having that attack "dip" is Molot in Sigma, but it's much less pronounced

Voxengo's Deft and Crunchessor can do it too + more curves.

[Ploki]

Push/Pump however is pretty unique.
The only ITB compressor that i saw do the "pump" attack phase as aggressively is the Fuse Audio VCL-515.
The only other compressor i've seen having that attack "dip" is Molot in Sigma, but it's much less pronounced

Voxengo's Deft and Crunchessor can do it too + more curves.

I actually have both how could i forgot.

You're right, especially Deft compressor can pump a lot. It also has fully parametric sidechain eq, and much lower CPU usage.

[jtsterays]

Push/Pump however is pretty unique.
The only ITB compressor that i saw do the "pump" attack phase as aggressively is the Fuse Audio VCL-515.
The only other compressor i've seen having that attack "dip" is Molot in Sigma, but it's much less pronounced

Voxengo's Deft and Crunchessor can do it too + more curves.

I actually have both how could i forgot.

You're right, especially Deft compressor can pump a lot. It also has fully parametric sidechain eq, and much lower CPU usage.

Its funny because you got me into those Voxengo comps in the 2023 compressors thread, such underated gems, crucial for shaping kicks/snares imo, ive been looking for those types of attack shapes since forever.

[Ploki]

Push/Pump however is pretty unique.
The only ITB compressor that i saw do the "pump" attack phase as aggressively is the Fuse Audio VCL-515.
The only other compressor i've seen having that attack "dip" is Molot in Sigma, but it's much less pronounced

Voxengo's Deft and Crunchessor can do it too + more curves.

I actually have both how could i forgot.

You're right, especially Deft compressor can pump a lot. It also has fully parametric sidechain eq, and much lower CPU usage.

Its funny because you got me into those Voxengo comps in the 2023 compressors thread, such underated gems, crucial for shaping kicks/snares imo, ive been looking for those types of attack shapes since forever.

I use Deft on snares all the time and Marquis on kicks, i just never bothered to analyze them i guess
Crunchessor is my fav guitar comp as well

[Agreed]

Big fan of Deft Compressor! Bought it in 2012, he's kept it up to date since quite impressively and it's always sounded great to me.

[bmanic]

The digital compressor clicks are from the peak detector not being being able to detect the peak (and thus the rms if rms detectors) of the signal in single sample rates. They have to emulate, well try to perform, the transform of the da converter to obtain the true peaks. 2-4x upsampling alone is not enough for many signals. See ITU BS.1770 paper with its estimated overshoot at various sample rates, which seems to be still off given the ice picks I have heard from every 1176 emulation. Thus without true peak detection, they cannot determine the polarity of the signal and thus switch between attack and release. Distortion in the side chain will also make the multiplier do stupid things to the audio.

The TDR compressors do not click on insane at faster attack. Klanghelm dc8c3 at 8x oversampling doesn’t either if set to smash and enough feedback. PSP Impressor didn’t either with the 16x oversampled sidechain.

Some of the processors with short lookaheads do not click too. The Softube Weiss DS-1 only clicks when you set attack and release so fast as to send it into oscillation or cause weird discontinuities in the switching between them. Oxford Dynamics (20 samples) and Pro C2 with a fraction of a millisecond of lookahead don’t either and can go pretty fast but they reach a point where they cannot speed up to my ears, Oxford Dynamics on release and Pro C2 on attack. The Oxford Dynamics can get the smoothness you mentioned along with Pro Audio DSP DSM, MDWDRC2, and Sound Radix Powair. The issue is lookahead slows down the rate of maximum modulation of the compressor by ramping the attack (or having a hold), prevents the compressor from compressing high frequencies because it modulates slower than their cycles unless the lookahead is very short like the Weiss DS-1’s default of 0.02 ms.

Analog can click too btw. The first transient to cross threshold of a dug in too much feedback compressor will often click horribly because its reacting to its own processing of the signal rather than the signal itself. A deeply dug in 1176 will misfire. Feedforward compressors can often leave plastic transients from lack of program dependency in poorer designed compressors but at least they react to the signal. DBX 160 series are some of the worst offenders because they expand when crossing threshold and then clamp down based on the sort of RMS with a slow attack and fast release unless you overcompress. They usually need to be limited afterward ime but at least RMS detectors are lower distortion than peak detectors. The same with the peak compressors with an attack curve that lets most of the initial peak pass through without even being shaped.

You are still describing the "wrong type of click". I'm not talking about ultra fast attack settings (though you can sometimes hear that clickyness there too). I'm talking about "chunky" settings, usually in the 5 to 30ms range. In the analogue domain there is very rarely a sort of "annoying clicky" sound to this type of compression whereas in the digital domain it is almost always present in some form. The biggest drawback with this is that the clickyness shows up once you keep processing the audio further. This is why I think it's an issue of aliasing because there too it usually gets worse the further you process a sound down the line.

A great example is the API 2500 compressor which is a master of "chunky" slamming (hence why it's so suitable for drumbus processing). Nobody has been able to a "perfect" emulation of this box yet, at least to my ears.. except maybe the DMG Audio Trackcomp one but only if you run it at stupid sample rates, meaning 192kHz session with absolutely maximum built in oversampling in the plugin. Then you get the attack sounding somewhat correct compared to the real deal. But for instance the waves version or the Lindel version sound wrong in the initial attack.

On a tangent: The best way to hear differences with AD/DA converters is to post process files, meaning after the AD/DA loop, add some saturation, some limiting etc. Once you've done some stuff to the audio you can much easier hear the difference between converters, even if the processing is 100% identical.

I don't know why this is, but seemingly very subtle differences in direct A/B comparisons yield much bigger differences down the line.

As for TDR compressors. Yes, those are among the absolute best ever made. Of course Kotelnikov can't do "chunky" at all. It's way too smooth at all times. But TDR Molot is absolutely excellent, yet to my ears even this one isn't all that good at doing "chunky".. it's still rather smooth. You can make it somewhat clicky but it doesn't do DBX160 style "plop!" chunk or the API 2500 type of.. hmm, how to describe it? .. "chunky" chunk.

Anyhow, these are after all absolutely minor details and don't really have any meaning but I'm just trying to describe what in my opinion makes this latest compressor from TBTECH really unique. It has an immediately recognizable mood or character to it when you use the de-click parameter in moderation.. and I have NOT been able to reproduce what it can do with anything else I have, which makes it 100% unique and thus very valuable in my opinion.

As some of you may know, I'm an absolute compressor junky and have virtually all of the plugin compressors on the market and have had a huge range of hardware compressors go through the various studios I've worked in (I myself am now virtually 100% ITB when it comes to mixing.. still got my Sintefex FX8000 here at home though). So yeah, I while I may be imagining things sometimes, with compressors I'm quite confident in what I am hearing.

I'll try to post some more sound examples of "chunky" compression. It's not the kind you use all that often but it's usually very good type of setting for drums and sometimes even bass. It's a great way to shape a sound, even if you end up flattening the source later down the pipeline.

[bmanic]

Push/Pump however is pretty unique.
The only ITB compressor that i saw do the "pump" attack phase as aggressively is the Fuse Audio VCL-515.
The only other compressor i've seen having that attack "dip" is Molot in Sigma, but it's much less pronounced

Voxengo's Deft and Crunchessor can do it too + more curves.

Voxengo's compressors are ALL absolutely must have and totally unique sounding. Heck, I STILL have the original Marquis installed on my studio computer because of it's unique compression action (and unfortunately you can't match the action with v2).

And Crunchessor is one of the few compressors that can do some really weird "chunky" compression in a completely unique way. Unfortunately, it's one of those that needs extreme oversampling to get somewhat rid of the bad clickyness. Still, it's very unique sounding and can sometimes work wonders on sources that just need some strange shaping.

As for Deft, it's a real workhorse compressor that sometimes can fool you due to the way the gain reduction meter works.