Quantcast
Cytomic "The Glue" Bus Compressor Effect Plugin - Page 91 - Gearspace.com
The No.1 Website for Pro Audio
Cytomic "The Glue" Bus Compressor Effect Plugin
Old 29th September 2017
  #2701
Lives for gear
 
🎧 5 years
Omg HD mod I’ll pay for the upgrade any guesstimate on time?
Old 29th September 2017 | Show parent
  #2702
Lives for gear
 
Marando's Avatar
 
2 Reviews written
🎧 10 years
Quote:
Originally Posted by andy-cytomic ➡️
The Range knob controls the voltage power supply levels in the sidechain section of The Glue, but I've calibrated it roughly in dB to make it more useful to customers. I noticed when modelling various G-Series compressors that the voltage level (ie how loud the signal can go) in the sidechain section changed the sound of the compressor when being pushed hard (low threshold). Their older gear ran off a lower voltage level than their newer gear. The threshold boosts the gain of the signal going into the sidechain section, and then the ratio shaper gains this up even more and is the first place clipping at the power voltage rails occurs. This is especially evident for transients, with low threshold and high ratio. Clipping the signal here stops compression from happening, it is prior to the envelope follower and makes the follower think the signal is quieter than it is so doesn't compress as much.

This is the same thing UAD copied off The Glue with their HR knob. But don't take my word for it, please go ahead and ask the guys at Universal Audio exactly what the HR knob does and confirm it directly from them if it controls the power voltage rail levels in the modelled device and causes clipping to occur, especially in the sidechain section with low threshold, and more specifically in the ratio shaper (which will be the first place to clip), and then post back here with what they say

edit: if they want to they can post directly to this thread! I'm all up for an open discussion
Thanks Andy for your explanation, much appreciated! It looks like the range knob in The Glue is doing something else than I though it was doing, so my bad!

I'm currently enjoying the new UA SSL compressor, but I have faith that your new and improved The Glue eventually will raise the bar even further. This faith is based on my experience with The Drop and The Scream, which - to me - are in a league of their own!
Old 29th September 2017
  #2703
Gear Addict
 
🎧 10 years
Andy, regarding newer, more powerful CPUs, and therefore new possibilities for you as a programmer: Will you be using stuff like AVX512, you reckon?
Old 30th September 2017 | Show parent
  #2704
Cytomic
 
andy-cytomic's Avatar
 
🎧 10 years
Quote:
Originally Posted by Melvin III ➡️
Omg HD mod I’ll pay for the upgrade any guesstimate on time?
I'm hoping to get something out by early Dec at the latest
Old 30th September 2017 | Show parent
  #2705
Cytomic
 
andy-cytomic's Avatar
 
🎧 10 years
Quote:
Originally Posted by Skap ➡️
Andy, regarding newer, more powerful CPUs, and therefore new possibilities for you as a programmer: Will you be using stuff like AVX512, you reckon?
In DSP SSE2 (128-bit) already covers pretty much everything needed, but there are some speedups to be had from AVX 512 (512-bit). I use lots of 128 bit operations on 4 x float 32-bit, and with 512 bits I can do 16 at once instead, which is considerable benefit. The main computation of the matrix inversion can't really benefit much from the 512-bits unless you want to run 8 channel audio instead of stereo!

(edit: I use as much SSE as possible to make as many operations as parallel as possible, whenever I see a plugin say "all double precision computations" I instantly think the developer is either using bad algorithms that need double precision as a bandaid, or they are lazy and can't be bothered working out where double precision is required and use float for everything else to make things run faster!)
Old 30th September 2017 | Show parent
  #2706
Gear Maniac
 
🎧 5 years
Quote:
Originally Posted by andy-cytomic ➡️
In DSP SSE2 (128-bit) already covers pretty much everything needed, but there are some speedups to be had from AVX 512 (512-bit). I use lots of 128 bit operations on 4 x float 32-bit, and with 512 bits I can do 16 at once instead, which is considerable benefit. The main computation of the matrix inversion can't really benefit much from the 512-bits unless you want to run 8 channel audio instead of stereo!

(edit: I use as much SSE as possible to make as many operations as parallel as possible, whenever I see a plugin say "all double precision computations" I instantly think the developer is either using bad algorithms that need double precision as a bandaid, or they are lazy and can't be bothered working out where double precision is required and use float for everything else to make things run faster!)
Hi Andy, hope you're well.

Half the reason I come to this site is to read all the technical mumbo jumbo like this. It's extremely fascinating to read, IMO. I wish more developers were active on here like you are, if only so I can geek out like crazy.
Old 30th September 2017 | Show parent
  #2707
Gear Addict
 
🎧 10 years
Quote:
Originally Posted by andy-cytomic ➡️
In DSP SSE2 (128-bit) already covers pretty much everything needed, but there are some speedups to be had from AVX 512 (512-bit). I use lots of 128 bit operations on 4 x float 32-bit, and with 512 bits I can do 16 at once instead, which is considerable benefit. The main computation of the matrix inversion can't really benefit much from the 512-bits unless you want to run 8 channel audio instead of stereo!

(edit: I use as much SSE as possible to make as many operations as parallel as possible, whenever I see a plugin say "all double precision computations" I instantly think the developer is either using bad algorithms that need double precision as a bandaid, or they are lazy and can't be bothered working out where double precision is required and use float for everything else to make things run faster!)
Thanks, Andy. I really appreciate it. You're one of my favorite developers. Always commited to the highest quality in every aspect of your business.

So would you say a plugin with a partly parallel path like f.i. multi band processor or perhaps a reverb with separate engines (2 for early reflections, 2 or more for reverb tail and other engines, like f.i. M7 and TC 6000 do) could possibly benefit from AVX512? If so, is it likely that more and more audio application/plugin developers will begin to optimize for AVX512 in addition to AVX2 and other instruction sets?

An AVX512 capable CPU (finally available in Intel's prosumer range) seems like a powerful feature for video, and I am wondering whether or not we will see any possible gains for audio use as well.

Thanks again, Andy.
Old 1st October 2017 | Show parent
  #2708
Cytomic
 
andy-cytomic's Avatar
 
🎧 10 years
Quote:
Originally Posted by deathbyguitar ➡️
Hi Andy, hope you're well.

Half the reason I come to this site is to read all the technical mumbo jumbo like this. It's extremely fascinating to read, IMO. I wish more developers were active on here like you are, if only so I can geek out like crazy.
Thanks! Hopefully the explanations, although technical are still approachable, but please ask away if there are some details I need to go into further
Old 1st October 2017
  #2709
Gear Maniac
 
Oni.'s Avatar
 
1 Review written
🎧 5 years
As for the fact that better CPUs = more precise emulations, there's someone that doesn't agree with us.



Last edited by Oni.; 1st October 2017 at 09:05 PM.. Reason: link not working
Old 2nd October 2017 | Show parent
  #2710
Cytomic
 
andy-cytomic's Avatar
 
🎧 10 years
Quote:
Originally Posted by Oni. ➡️
As for the fact that better CPUs = more precise emulations, there's someone that doesn't agree with us.
Well I'm happy to lead the way and prove him wrong The Scream is a start in that direction and there will be plenty more plugins like it to follow.

I take inspiration from the 3D world where special tricks are losing in terms of the best looking renders. Time and time again more physically accurate optimised brute forced solutions are used. If you skip through you'll hear this time and time again in a talk from a guru in optimised 3D rendering John Karmack (quake): https://youtu.be/P6UKhR0T6cs
Old 2nd October 2017
  #2711
Lives for gear
 
🎧 10 years
Ray tracing is very different to digital signal processing. Right? They are completely different fields of computer science.
Old 2nd October 2017 | Show parent
  #2712
Deleted bc6012f
Guest
Quote:
Originally Posted by miscend ➡️
Ray tracing is very different to digital signal processing. Right? They are completely different fields of computer science.
Of course they are. We talk about these subjects here in metaphor relations ie to better understand overall subject in the long run.
Old 2nd October 2017 | Show parent
  #2713
Lives for gear
 
🎧 10 years
Quote:
Originally Posted by Deleted bc6012f ➡️
Of course they are. We talk about these subjects here in metaphor relations ie to better understand overall subject in the long run.
Ray tracing is an offline process rendered over several days using huge render farms. They actively sacrife efficiency for accuracy and it requires massive amounts of computing power. DSP is about real time applications with minimum latency.

A closer metaphor to the DSP would be rasterisation. Which is the real time rendering of 3D graphics. Using tricks like global illumination and downsampling (also called Ordered Grid SuperSampling AntiAliasing) you can get very close to ray traced quality images. And as GPUs get ever more powerful and the techniques they use to do the tricks get more advanced you practically won’t be able to the difference.
Old 3rd October 2017 | Show parent
  #2714
Cytomic
 
andy-cytomic's Avatar
 
🎧 10 years
Quote:
Originally Posted by miscend ➡️
Ray tracing is an offline process rendered over several days using huge render farms. They actively sacrife efficiency for accuracy and it requires massive amounts of computing power. DSP is about real time applications with minimum latency.

A closer metaphor to the DSP would be rasterisation. Which is the real time rendering of 3D graphics. Using tricks like global illumination and downsampling (also called Ordered Grid SuperSampling AntiAliasing) you can get very close to ray traced quality images. And as GPUs get ever more powerful and the techniques they use to do the tricks get more advanced you practically won’t be able to the difference.
I think you've missed a massive chunk of what I'm talking about here. Only realtime DSP is about realtime applications with minimum latency. What happens when you click "freeze" or "render" or "bounce"? This is analogous to the world of 3D rendering where you have a realtime preview of your scene, but no one seriously expects this to be what you see at the movies! Why should this be any different in the audio world?!!!

This is what I'm pushing for here and the entire point of my posts. The realtime and render worlds of 3D can easily be applied to audio. Wouldn't you like to be able to print your stems through the worlds most expensive analog equipment if it took a few hours of waiting for the advanced rendering? Who cares if it takes a bit longer, just print the song overnight and come back to the most gorgeous sounding non-linear analog modelled processing possible with no aliasing. In fact we can even get better results than analog since none of the transistors produce any hiss.

As for the realtime audio DSP vs realtime games / preview 3D world, then even here please actually look at John Carmack's talk about where things are heading. It's all about physically accurate optimised brute force models winning over scan line hacks. Here is that link again so you can't miss it: https://youtu.be/IyUgHPs86XM

Also a point on the overlap between 3D and audio. Although 3D is inherently more parallel that audio there is still a huge amount of overlap in terms of optimisation and functional approximations. Oversampling, anti-aliasing, fast inverse (1/x), fast inverse sqrt, fast matrix operations and linear algebra - there is a huge amount of overlap in these basic number crunching tasks. There is also a huge overlap in finite element analysis and simulation of physical bodies, so non-linear sparse matrix solving. Here is a course from MIT about "computation maths" by the wonderful professor Gilbert Strang https://ocw.mit.edu/courses/mathemat...g-i-fall-2008/ in which he goes through the matrix equations for numerically solving both linear and non-linear differential equations that arise in all aspects of engineering (eg circuits for audio, mass springs, vibrational bodies etc) and how the equations and techniques are the same. Sure the 3D world comes at things from a different angle, but it's all the same type of stuff under the hood - hardcore number crunching.

Last edited by andy-cytomic; 4th October 2017 at 02:10 AM.. Reason: fixed typo
Old 3rd October 2017 | Show parent
  #2715
Gear Maniac
 
🎧 5 years
Quote:
Originally Posted by andy-cytomic ➡️
Wouldn't you like to be able to print your stems through the worlds most expensive analog equipment if it took a few hours of waiting for the advanced rendering?
Yes. Yes I would. Is this ultimately the goal for The Glue going forward? To be able to rival if not surpass the quality of compressor hardware by using super-detailed offline rendering? That would be so awesome.

Is there still something "magical" to be gained from a real SSL 4k unit (or a clone, for that matter) vs an emulation? Some of my favorite in-the-box producers are using hardware comps for their master bus these days and I wonder if we're just a little ways off from "the real thing" being completely matched by software/digital solutions. Fractal Audio sure seems to be doing that with guitar amps, at least.
Old 4th October 2017 | Show parent
  #2716
Cytomic
 
andy-cytomic's Avatar
 
🎧 10 years
Quote:
Originally Posted by deathbyguitar ➡️
Yes. Yes I would. Is this ultimately the goal for The Glue going forward? To be able to rival if not surpass the quality of compressor hardware by using super-detailed offline rendering? That would be so awesome.

Is there still something "magical" to be gained from a real SSL 4k unit (or a clone, for that matter) vs an emulation? Some of my favorite in-the-box producers are using hardware comps for their master bus these days and I wonder if we're just a little ways off from "the real thing" being completely matched by software/digital solutions. Fractal Audio sure seems to be doing that with guitar amps, at least.
Actual SSL 4000 bus compressors or any of the rack versions or clones will definitely sound better than any SSL bus compressor plugin I have heard, including the 9 year old Glue! I hope to change that with the HD update of The Glue

I've not heard any of the Fractal Audio stuff yet, so thanks for pointing them out. From their web page it sounds like they are on the right track for the pre amp and power amp sections. They are still using IRs for the cab/speaker/room/mic, so a big chunk of the "real" tone will be missing. A full power amp model coupled to a physical speaker/cab model is the way forwards here, but that is computationally very expensive and probably still a way off yet in terms of reasonable cost. IRs are ok for now till compute power and algorithms both improve.
Old 4th October 2017 | Show parent
  #2717
Gear Maniac
 
🎧 5 years
Quote:
Originally Posted by andy-cytomic ➡️
Actual SSL 4000 bus compressors or any of the rack versions or clones will definitely sound better than any SSL bus compressor plugin I have heard, including the 9 year old Glue! I hope to change that with the HD update of The Glue

I've not heard any of the Fractal Audio stuff yet, so thanks for pointing them out. From their web page it sounds like they are on the right track for the pre amp and power amp sections. They are still using IRs for the cab/speaker/room/mic, so a big chunk of the "real" tone will be missing. A full power amp model coupled to a physical speaker/cab model is the way forwards here, but that is computationally very expensive and probably still a way off yet in terms of reasonable cost. IRs are ok for now till compute power and algorithms both improve.
I'm a bit of a bit of a Fractal fanboy so forgive my bias, but I'd argue that with the detail isn't missing at all in their implementations. They model amps and effects down to the component level (Yeah, I know everyone says this) and the results are often indistinguishable from the real thing. They fully model the preamp and power amp sections of all amps. The computationally expensive thing really isn't that big of an issue when you have dual TigerSHARC DSPs to play around with along with world-class coverters and blah blah blah. It's why the units cost so much. I HIGHLY recommend you read up on the Axe-FX II and Fractal's approach to modeling. There's a reason why many of top artists/bands have ditched their real amps and just play Fractals now.

Old 4th October 2017 | Show parent
  #2718
Cytomic
 
andy-cytomic's Avatar
 
🎧 10 years
Quote:
Originally Posted by deathbyguitar ➡️
I'm a bit of a bit of a Fractal fanboy so forgive my bias, but I'd argue that with the detail isn't missing at all in their implementations. They model amps and effects down to the component level (Yeah, I know everyone says this) and the results are often indistinguishable from the real thing. They fully model the preamp and power amp sections of all amps. The computationally expensive thing really isn't that big of an issue when you have dual TigerSHARC DSPs to play around with along with world-class coverters and blah blah blah. It's why the units cost so much. I HIGHLY recommend you read up on the Axe-FX II and Fractal's approach to modeling. There's a reason why many of top artists/bands have ditched their real amps and just play Fractals now.
Please re-read my post. The chunk that I was talking about missing is in fact nothing that you have mentioned. I'm talking about specifically the IR section, which is all the stuff post power amp that will most likely be better in the future, but right now it's an ok compromise.

Two TigerSHARC DSP's is fine for doing some pedals and pre and power amp in full detail, no worries, and also tack on some IRs to get a snapshot linear lumped model of "everything else". A detailed model of the coupling of power amp with a physical model of the speaker and coil and cabinet and room and mic is computationally very expensive and still a way off, but will probably be done in the future since there is more tone to be had from all the non-linearities there.
Old 4th October 2017 | Show parent
  #2719
Gear Maniac
 
🎧 5 years
Quote:
Originally Posted by andy-cytomic ➡️
Please re-read my post. The chunk that I was talking about missing is in fact nothing that you have mentioned. I'm talking about specifically the IR section, which is all the stuff post power amp that will most likely be better in the future, but right now it's an ok compromise.

Two TigerSHARC DSP's is fine for doing some pedals and pre and power amp in full detail, no worries, and also tack on some IRs to get a snapshot linear lumped model of "everything else". A detailed model of the coupling of power amp with a physical model of the speaker and coil and cabinet and room and mic is computationally very expensive and still a way off, but will probably be done in the future since there is more tone to be had from all the non-linearities there.
Fair enough. I'm just a music tech enthusiast, not an engineer of sorts. Thanks for the insight.
Old 4th October 2017 | Show parent
  #2720
Lives for gear
 
🎧 10 years
Quote:
Originally Posted by andy-cytomic ➡️
Actual SSL 4000 bus compressors or any of the rack versions or clones will definitely sound better than any SSL bus compressor plugin I have heard, including the 9 year old Glue! I hope to change that with the HD update of The Glue
Forgive me if this is a dumb question, but if we leave aside factors such as aliasing, accuracy (or lack thereof) of modeled saturation, and any other short comings of plugins when compared to real analog gear, are the current crop of compressor plugins also coming up short simply in terms of their ability to control dynamics?
I ask because many people are raving about Slate's FG-Stress being somehow groundbreaking in its attack and release. Similar things are being said about the new Kush Novatron.
Will advancements in compressor plugins change the way they control volume (for lack of a better word) or will the changes be confined to lower aliasing, more accurate saturation, etc.
Old 4th October 2017 | Show parent
  #2721
Gear Maniac
 
Oni.'s Avatar
 
1 Review written
🎧 5 years
Quote:
Originally Posted by Squeegee 303 ➡️
Forgive me if this is a dumb question, but if we leave aside factors such as aliasing, accuracy (or lack thereof) of modeled saturation, and any other short comings of plugins when compared to real analog gear, are the current crop of compressor plugins also coming up short simply in terms of their ability to control dynamics?
I ask because many people are raving about Slate's FG-Stress being somehow groundbreaking in its attack and release. Similar things are being said about the new Kush Novatron.
Will advancements in compressor plugins change the way they control volume (for lack of a better word) or will the changes be confined to lower aliasing, more accurate saturation, etc.
Attack & release times are 50% of where the money is, of course devs are gonna implement more precise timing in the future.
Old 4th October 2017
  #2722
Gear Guru
 
jwh1192's Avatar
 
2 Reviews written
🎧 15 years
patiently waiting for the HD version of the Glue .. thx Andy for continuing to make this plugin even better !!!
Old 4th October 2017
  #2723
Gear Guru
 
4 Reviews written
🎧 10 years
Andy is one of the good guys! Constantly contributing in a positive way to GS. No hype just substance.
Old 5th October 2017 | Show parent
  #2724
Gear Addict
 
🎧 10 years
Quote:
Originally Posted by Oni. ➡️
Attack & release times are 50% of where the money is, of course devs are gonna implement more precise timing in the future.
I'm pretty certain Andy did those 50% right in The Glue. I did some tests with the help of the friends with actual HW SSLs and at least as far as attack and release envelopes go, The Glue is spot on.

Now for the rant...
It's interesting to ponder -- what's left in there?
Is it only saturation? How much is it important? If an absolutely accurate idealized "compressor" is made in digital (I believe it is possible, or at least close to being possible in DSP these days), should it necessarily sound bland? Would it be too perfect?
If a precise model of a hardware is realized in digital wouldn't it include compromises made during the design/manufacturing?
Are there improvements to be made over hardware?
I've kinda grown tired of emulations. Between Waves Torque and Slate's FG-Stress, I'm more excited about the former as it does something nothing else in my disposal even remotely does. Don't get me wrong, FG-Stress is very cool! I'd love to see either something totally unique or hardware modeling which improves upon original (leaving an option to be true to prototype, of course). The Scream is great in that respect!
...rant end

As far as The Glue HD, recent Plugin-Alliance releases got me thinking, maybe component level mismatches do something important, e.g. provide some decorrelation between L and R channels which opens stereo field?
Andy, do you plan to implement or at lest look into something like that?
Old 6th October 2017 | Show parent
  #2725
Gear Maniac
 
Oni.'s Avatar
 
1 Review written
🎧 5 years
Quote:
Originally Posted by meloco_go ➡️
I'm pretty certain Andy did those 50% right in The Glue. I did some tests with the help of the friends with actual HW SSLs and at least as far as attack and release envelopes go, The Glue is spot on.

Are there improvements to be made over hardware?
Actually, not to specifically criticise your point, but I think nowadays people are exagerating a bit their expectation in terms of saturation: yes it can sound pleasing when done right, but that's not everything. I still have to hear a digital compressor shaping envelopes the way an analog one does ( outside of the UAD world. Those guys really got it right ).

Quote:
Originally Posted by meloco_go ➡️
Now for the rant...
It's interesting to ponder -- what's left in there?
Is it only saturation? How much is it important? If an absolutely accurate idealized "compressor" is made in digital (I believe it is possible, or at least close to being possible in DSP these days), should it necessarily sound bland? Would it be too perfect?
If a precise model of a hardware is realized in digital wouldn't it include compromises made during the design/manufacturing?
As for your rant.. there are some pretty top-end mastering compressors that are digital. One of Bob Katz's go-to compressors is a fully digital one. It only exsists in outboard format, but it's 100% digital. There are simply clean compressors and colored compressors, and all the shades in between, and considering a unit "bland" because it doesn't impart "color" on the processed signal is really an error.
Old 6th October 2017 | Show parent
  #2726
Cytomic
 
andy-cytomic's Avatar
 
🎧 10 years
Quote:
Originally Posted by Squeegee 303 ➡️
Forgive me if this is a dumb question, but if we leave aside factors such as aliasing, accuracy (or lack thereof) of modeled saturation, and any other short comings of plugins when compared to real analog gear, are the current crop of compressor plugins also coming up short simply in terms of their ability to control dynamics?
I ask because many people are raving about Slate's FG-Stress being somehow groundbreaking in its attack and release. Similar things are being said about the new Kush Novatron.
Will advancements in compressor plugins change the way they control volume (for lack of a better word) or will the changes be confined to lower aliasing, more accurate saturation, etc.
I don't think there will be any fundamentally better sounding processing of volume than we currently have, just different flavours. Probably one area that may improve is more "auto" gain reduction, like photographers have with their cameras there are a bunch of algorithms under the hood analysing the image and adjusting it to get a good photo, this sort of thing will most likely happen with audio to speed up mixing.

The Glue is an analog model of a particular bit of hardware, so there isn't much room in changing how the action of the compression works, it's a diode based one pole envelope follower with twin VCA amps.

I'm a particular fan of building a continuous time non-linear system and then solving it in detail, and this is at the heart of analog modelling but doesn't preclude doing things not possible in circuits. There are certain limits to smoothness vs harmonic generation possible, there is always a tradeoff in these things in both analog and digital, it's just a matter of deciding what sounds best.
Old 6th October 2017 | Show parent
  #2727
Cytomic
 
andy-cytomic's Avatar
 
🎧 10 years
Quote:
Originally Posted by meloco_go ➡️
I'm pretty certain Andy did those 50% right in The Glue. I did some tests with the help of the friends with actual HW SSLs and at least as far as attack and release envelopes go, The Glue is spot on.

Now for the rant...
It's interesting to ponder -- what's left in there?
Is it only saturation? How much is it important? If an absolutely accurate idealized "compressor" is made in digital (I believe it is possible, or at least close to being possible in DSP these days), should it necessarily sound bland? Would it be too perfect?
If a precise model of a hardware is realized in digital wouldn't it include compromises made during the design/manufacturing?
Are there improvements to be made over hardware?
I've kinda grown tired of emulations. Between Waves Torque and Slate's FG-Stress, I'm more excited about the former as it does something nothing else in my disposal even remotely does. Don't get me wrong, FG-Stress is very cool! I'd love to see either something totally unique or hardware modeling which improves upon original (leaving an option to be true to prototype, of course). The Scream is great in that respect!
...rant end

As far as The Glue HD, recent Plugin-Alliance releases got me thinking, maybe component level mismatches do something important, e.g. provide some decorrelation between L and R channels which opens stereo field?
Andy, do you plan to implement or at lest look into something like that?
I've already component mismatches in mono and stereo. Stereo mismatches help a little, but should be optional. I've already coded and just about to release mismatches in The Scream, where you can set a level of tolerance in both mono and stereo and adjust the depth of mismatch with a fader to taste. There is a random seed that you use, which is saved in your preset, so you can that variation easily with friends. This way you have complete control over the variations. You can also generate a new random seed to base the variations from, or manually enter your own.

The Glue is an idealised compressor, only the non-linear envelope follower and sidechain are modelled, but they are required to make it a compressor in the first place! The optional HD mode will make it sound closer to hardware by modelling the full non-linear signal chain, and in particular a detailed model of the internals of the VCA.

Precise models can optionally include anything. In The Scream you can adjust all parameters of all models, including bypassing things you don't want. For example the finite beta effects of transistors can be bypassed completely if you want, but normally these small imperfections add to the tone so you want to keep them.

Probably the main thing you may want to bypass, or have control of the level, is noise. Analog electronics have thermal noise, shot noise, flicker noise, and burst noise to name a few. Modelling these realistically takes a fair chunk of cpu, and is something I'm interested in tackling in the future as I feel this is the last piece of the puzzle in terms of tone.

I will not be able to model things like RF interference from local radio stations since the plugin has no physical components to act as an antenna, but I'm pretty sure most people don't need this in their analog models!
Old 6th October 2017 | Show parent
  #2728
Cytomic
 
andy-cytomic's Avatar
 
🎧 10 years
Quote:
Originally Posted by Oni. ➡️
Actually, not to specifically criticise your point, but I think nowadays people are exagerating a bit their expectation in terms of saturation: yes it can sound pleasing when done right, but that's not everything. I still have to hear a digital compressor shaping envelopes the way an analog one does ( outside of the UAD world. Those guys really got it right ).
The Glue has an accurate one pole low pass diode based envelope follower that matches the analog SSL G-Series compressor. This naturally has super low aliasing because of the physics of the circuit that is accurately captured in The Glue, which is a big part of why it sounds so good.

The Glue's harmonic spectrum generated by the envelope follower has way less aliasing than the original UAD SSL bus compressor model, please view back in this thread for plots, or check the "ultimate plugin analysis thread" plots eg: https://gearspace.com/board/5445398-post498.html

These plots (not done by me) use an old version of The Glue that had a perfectly symmetric ratio shaper section (so have no even harmonics), but this will not impact the aliasing levels due to the accurate modelling of the envelope follower action. If you browse the thread you will see more plots showing the even harmonics and still the lowest aliasing of any of the SSL compressor models.
Old 6th October 2017 | Show parent
  #2729
Lives for gear
 
🎧 10 years
Quote:
Originally Posted by andy-cytomic ➡️
I will not be able to model things like RF interference from local radio stations since the plugin has no physical components to act as an antenna, but I'm pretty sure most people don't need this in their analog models!
Without true RF interference Spinal Tap solos ITB are not possible. This may be for the best though.
Old 6th October 2017 | Show parent
  #2730
Gear Addict
 
🎧 10 years
Quote:
Originally Posted by andy-cytomic ➡️
I've already component mismatches in mono and stereo. Stereo mismatches help a little, but should be optional. I've already coded and just about to release mismatches in The Scream, where you can set a level of tolerance in both mono and stereo and adjust the depth of mismatch with a fader to taste. There is a random seed that you use, which is saved in your preset, so you can that variation easily with friends. This way you have complete control over the variations. You can also generate a new random seed to base the variations from, or manually enter your own.
That is seriously cool! It would be really interesting to see how it pans out in The Scream, and what it would do in The Glue HD.
📝 Reply

Similar Threads

Thread / Thread Starter Replies / Views Last Post
replies: 4616 views: 615429
Avatar for smoke
smoke 7th May 2021
replies: 98 views: 39199
Avatar for dfghdhr
dfghdhr 5th June 2021
replies: 295 views: 74343
Avatar for anguswoodhead
anguswoodhead 26th March 2013
replies: 1296 views: 181830
Avatar for heraldo_jones
heraldo_jones 1st February 2016
Post Reply

Welcome to the Gearspace Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…

Forum Jump
Forum Jump