Cytomic 'The Glue' Compressor

[andy-cytomic]

Sounds like great progress, Andy! With developments like sub-circuits coming online, how much more big R&D work do you think is still ahead before you can start really operationalizing all this (i.e. getting plugins out the door that leverage the new tech)? The R&D itself is super cool, but of course, it's ultimately a means to an end. I genuinely hope you're able to reach a point where circuit simulation becomes almost trivial on your end, and it allows you to turn around insanely high-quality plugins, with The Scream-level attention to detail, at a faster pace than before. And I don't say that as a knock or criticism, just more of a wish for a more Cytomic yellow and black future.

I think despite claims from lots of vendors, very few are doing analog emulation at anywhere near the levels you're on the cusp of and it's exciting.

The most important part of the Glue R&D was getting the discrete Blackmer VCA model up and running, which is done. The R&D left for The Glue v2 is implementing a JFET input op-amp model, which is this week's work, and then investigating a basic gyrator based non-linear transformer model, which I want to spend at most a month on. I've got a DIY-RE SSL bus compressor kit and a that I'm going to measure each component of, writing that value on the schematic, and then assemble, which will save some guesswork when matching to my model. It has an optional transformer output stage, which I'll use to see how good the gyrator transformer model is, and either include it if I'm happy, or leave it out for a later update if I'm not happy with the sound of the model compared to the analog device. I've also got a Stam audio SA-4000 mk3 to compare to, but they wouldn't provide schematics, so it may be difficult to match any presets to that if they've altered the circuit too much.

Then it's all the UI and application level stuff to do, and all the regular packaging stuff like manual, audio examples, videos, and presets. I will report progress on all of these steps as I do them, and it will take as long as it takes.

The DSP output of the Cytomic Circuit Solver at any point in time is always production ready. For testing I skip the extended optimisations, but at any time for a release I switch that back on and it takes around an hour to crunch through a bunch of automated symbolic optimisations to keep the cpu as low as possible.

What I am hoping for is that with the extra income from The Glue v2 I can hire someone to help with the UI and application C++ and web and marketing and demo / tutorial type stuff so I can concentrate on the circuit solver / DSP and then it will be much quicker to complete products since once the DSP is done the rest of the work can be done in parallel with me working on the next chunk product.

[Funkybot's Evil Twin]

Sounds great Andy. Thanks for the detailed response!

[meloco_go]

Andy, could you please consider, as a unique special feature, adding an option for The Glue to output and accept a GR signal?

For example, an instance of the plugin could output audio through outs 1-2 and the GR signal through outs 3-4. Likewise, an instance could accept a GR signal through inputs 3-4. Internally, GR could be summed or "voted" in a manner similar to how left-right GR is weighted.

This would enable several neat tricks, such as setting up a surround/Atmos mix (if the DAW allows for complex routing) or creating a complex "parallel" compression setup, where two instances react to different frequency ranges within a mix and share GR.
I imagine this could cause issues in certain DAWs, and possibly even damage speakers by sending a DC signal, so the setting should probably be hidden under an "expert" tab and include a disclaimer.

Alternatively, maybe there’s a way to tunnel signals between plugin instances without outputting them. I believe some plugins can do that.

[andy-cytomic]

Andy, could you please consider, as a unique special feature, adding an option for The Glue to output and accept a GR signal?

For example, an instance of the plugin could output audio through outs 1-2 and the GR signal through outs 3-4. Likewise, an instance could accept a GR signal through inputs 3-4. Internally, GR could be summed or "voted" in a manner similar to how left-right GR is weighted.

This would enable several neat tricks, such as setting up a surround/Atmos mix (if the DAW allows for complex routing) or creating a complex "parallel" compression setup, where two instances react to different frequency ranges within a mix and share GR.
I imagine this could cause issues in certain DAWs, and possibly even damage speakers by sending a DC signal, so the setting should probably be hidden under an "expert" tab and include a disclaimer.

Alternatively, maybe there’s a way to tunnel signals between plugin instances without outputting them. I believe some plugins can do that.

I think a dedicated surround version of The Glue would be better suited for the surround side of things, and that is something I've considered before to support the film industry.

I can see certain niche use cases for routing dc signals, for example in printing stems in a DAW like PT which only has mono sidechain routing, since The Glue needs a stereo sidechain to work properly. This is a pretty niche use case, and something way down the list of what I want to spend my time on, sorry.

[jtsterays]

Have you heard any significant sonic improvements with the added details? Or is this too early to ask?

[andy-cytomic]

Have you heard any significant sonic improvements with the added details? Or is this too early to ask?

I am confident it will sound more like an analog g-series bus compressor, so if that's a significant sonic improvement then yes - but the differences may be subtle in some cases. There will be some extra drive from various main signal path non-linearities, even without the compressor doing anything (and lots more if you clip the input op-amps), and the Blackmer VCA will add a little (or a lot if mismatched) drive and some control voltage bleed-through from the envelope follower will add a little extra thump. I think the main thing people will hear, if stereo randomisation is enabled, is a widening effect caused by slightly different gain reduction amounts being applied to the left and right channels, as all the little variations in component values cause subtle variations between the channels.

[meloco_go]

I can see certain niche use cases for routing dc signals, for example in printing stems in a DAW like PT which only has mono sidechain routing, since The Glue needs a stereo sidechain to work properly. This is a pretty niche use case, and something way down the list of what I want to spend my time on, sorry.

Yes, sure. Maybe one day when you get the thing up and running!

[andy-cytomic]

I can see certain niche use cases for routing dc signals, for example in printing stems in a DAW like PT which only has mono sidechain routing, since The Glue needs a stereo sidechain to work properly. This is a pretty niche use case, and something way down the list of what I want to spend my time on, sorry.

Yes, sure. Maybe one day when you get the thing up and running!

Is there something that I'm missing with regards the functionality provided by DC signal routing? From what I know, in all DAWs, apart from ProTools, you can already send stereo external sidechain signals to multiple copies of The Glue and achieve any compression task that you would otherwise do with dc-signal routing, and regular audio signals can be EQ'd or gated or soloed etc so to me this is more powerful and less confusing. What is the task that you could achieve via DC processing that you can't via regular external sidechain routing?

[meloco_go]

Is there something that I'm missing with regards the functionality provided by DC signal routing? From what I know, in all DAWs, apart from ProTools, you can already send stereo external sidechain signals to multiple copies of The Glue and achieve any compression task that you would otherwise do with dc-signal routing, and regular audio signals can be EQ'd or gated or soloed etc so to me this is more powerful and less confusing. What is the task that you could achieve via DC processing that you can't via regular external sidechain routing?

My idea will make sense if the GR signals are combined the same way stereo linking is done (highest GR wins). As far as I am aware, this is also the way the dynamics modules are linked on the console.

This approach enables techniques that are not possible otherwise and are different from simply sending sidechain audio.

For example, you could set up one instance of The Glue to process the full mix and another instance on an EQ'd version (such as a DeEsser with fast attack and release). The second instance shares its GR with the first, so if the second experiences a momentarily higher GR, the first will respond accordingly. Once the GR spike in the second instance subsides, the first returns to its own GR. This behavior is distinct from placing two instances in series. Granted, the difference may be small.

Also, while dedicated surround version is surely convenient, the "shared GR" approach is far more flexible.

[Tao Jones]

If I buy the Glue v1, would I be able to update to v2 at a discount? I'm really digging reading about your development of v2!
Oh and thanks for adding meter support for UC1! Not a lot of plugins support the UC1 meter.

[andy-cytomic]

If I buy the Glue v1, would I be able to update to v2 at a discount? I'm really digging reading about your development of v2!
Oh and thanks for adding meter support for UC1! Not a lot of plugins support the UC1 meter.

The Glue v2 will be a free update to all those that bought The Glue v1 from Cytomic for USD 99, and for those that got a discount off their purchase from Cytomic they can pay the difference up to USD 99 for the upgrade.

[andy-cytomic]

What would people prefer for The Glue v2:
1) v1 and v2 are different plugins
2) v2 breaks existing projects with new changes, but still loads in place of v1

If both versions are different plugins then to transition to v2 you will need to manually add v2 next to where v1 is, then manually copy and paste the preset. This will mean all old projects remain 100% sound identical, and all existing automation will work perfectly.

The other option is to break people's existing projects with The Glue v2 which will have more ratios to choose from, a different makeup gain range, and a few other things which will not be 100% compatible with the old version, so people's automation will break, and the sound will be different under certain circumstances.

Unfortunately there is no middle ground here, since I'm not a DAW so can't transition automation etc to new parameters / ranges.

[pough]

If it breaks old projects then it's a different plugin. That's how I see it, anyways.

[DubLifeSD]

I'd prefer a separate v2 plugin, or shall I say unglued version.

[Andreya_Autumn]

What would people prefer for The Glue v2:
1) v1 and v2 are different plugins
2) v2 breaks existing projects with new changes, but still loads in place of v1

I really like the way Fabfilter did the Pro-Q 3 to 4 update. They pushed an update to 3 that lets you copy instance state to the system clipboard, which can then be pasted into an instance of 4. Automation won't carry over that way, obviously. But still, this made updating old projects very quick and easy in my experience.