Fabfilter Pro-C 3

[E_Anderson]

Re-clicking the comp mode menu button to close it without selection re-opens the already-open menu if you move your mouse at all while the menu is open

Just submitted a reproducible crash for CLAP version in REAPER, enter the preset text-search field and then press esc, 100% crash rate

[oobesan]

If you just want to compress the signal without changing the level, auto gain is supposed to do that for you. I mean, why would anyone consider that to be a bad thing?

I gotta say, if I'm understanding you correctly I think I would consider it a bad thing. Please forgive and correct me if I misunderstood, but what I think you are asking for is an auto-gain feature that would change the makeup gain dynamically based on the input signal. That in essence is additional compression, though perhaps quite slow, that some might rather refer to as "leveling". The problem with that is you would effectively have no control of this additional compression as it's just an on-off switch for a mysterious unknown algorithm with a preset configuration working behind the scenes. If your input signal is not consistent enough throughout the entire program material, you would likely be dealing with changes in level that are unpredictable at best and likely undesirable. Countering that by automating or somehow normalizing the input would be cumbersome and confusing, and might even require additional compensating automation on the output. The alternative of course is calculating a static makeup gain based on the current settings, with that calculation having been calibrated on some test signal deemed to be as representative as possible of basically every possible signal anyone is likely to feed the compressor. That's obviously impossible to get right as the exact same settings are likely to require different makeup gain for different signals. It's therefore far from perfect, but considering the alternative "dynamic" option it's also in my opinion the better choice. You'll usually still have to adjust the makeup gain anyway but it might at least get you closer to begin with. As bmanic suggested, that's probably what FabFilter (and likely many others as well) are doing. I will say, I have encountered a third option with a button that will let you sample some representative section of the signal for a few seconds and then set the makeup gain dial to a (static) position calculated by preforming some level-matching algorithm on that snippet with the current settings. Fircomp comes to mind but there may be others. Not sure what Newfangled are doing but I'm about to go check it out!

So, all I am talking about is gain staging. Different compressors handle this differently. Some compressors have separate gain staging controls for input, make-up, and output. Logic Pro's Compressor is like this. I can drive it into the compressor hard with the input knob and have aggressive settings to squash the signal. Then, I can add makeup gain. But, this is before the limiter and distortion modules, and before the final output gain stage. Makeup gain and Output gain are two different gain stages.

UAD's "purity test" versions of the 1176 bluey has input and output, but no makeup. The classic SSL bus compressors have no input gain control (you would have to drive it with a separate gain utility), but have an output gain control that they call "makeup". Pro C 2 does not have input or make up, but do have an output. In reality, SSL bus comp makeup and Pro C 2's output are the same, and both are different from Logic Pro's make up control. Other compressors have bare-minimum controls. The One Knob plugins from Waves are like this. Just a drive (input) knob.

And, of course, some have an autogain feature. And some do that better than others.

Whatever you call the final gain stage (makeup, output, whatever), it's all just gain staging. And gain staging is just tedious, especially if you have a chain of plugins after the compressor. An autogain feature is a very welcome addition to handle this, when it's done well. Like, everyone on earth other than KVR aficionados feels this is a very useful option to have. You either have a good autogain function, or you deal with it manually. I don't know why anyone would enjoy gain staging manually rather than you know, push the Autogain button and move on with your mix. But maybe y'all have a perspective I have not considered.

Anyway... none of this changes the fact that Pro C has an Autogain feature, and it sucks, because it doesn't do what it says it does.

[E_Anderson]

A static implementation does what it says it does even if it's simply internally linked parameters based on assumed input strength.

Maybe your point isn't about the type of autogain used, but whether it works well, whatever method the developers go with. Personally I completely agree that I'd like improvement here, there's only upside to better autogain.

Maybe the best I've heard has been by PSP in Impressor. Very often, the default autogain value is spot-on and tracks the Compression knob very accurately. Actually, their implementation really outperforms FF/Presswerk/Cenozoix very noticeably (if memory serves - I've uninstalled Impressor).

[kraster]

If you just want to compress the signal without changing the level, auto gain is supposed to do that for you. I mean, why would anyone consider that to be a bad thing?

I gotta say, if I'm understanding you correctly I think I would consider it a bad thing. Please forgive and correct me if I misunderstood, but what I think you are asking for is an auto-gain feature that would change the makeup gain dynamically based on the input signal. That in essence is additional compression, though perhaps quite slow, that some might rather refer to as "leveling". The problem with that is you would effectively have no control of this additional compression as it's just an on-off switch for a mysterious unknown algorithm with a preset configuration working behind the scenes. If your input signal is not consistent enough throughout the entire program material, you would likely be dealing with changes in level that are unpredictable at best and likely undesirable. Countering that by automating or somehow normalizing the input would be cumbersome and confusing, and might even require additional compensating automation on the output. The alternative of course is calculating a static makeup gain based on the current settings, with that calculation having been calibrated on some test signal deemed to be as representative as possible of basically every possible signal anyone is likely to feed the compressor. That's obviously impossible to get right as the exact same settings are likely to require different makeup gain for different signals. It's therefore far from perfect, but considering the alternative "dynamic" option it's also in my opinion the better choice. You'll usually still have to adjust the makeup gain anyway but it might at least get you closer to begin with. As bmanic suggested, that's probably what FabFilter (and likely many others as well) are doing. I will say, I have encountered a third option with a button that will let you sample some representative section of the signal for a few seconds and then set the makeup gain dial to a (static) position calculated by preforming some level-matching algorithm on that snippet with the current settings. Fircomp comes to mind but there may be others. Not sure what Newfangled are doing but I'm about to go check it out!

So, all I am talking about is gain staging. Different compressors handle this differently. Some compressors have separate gain staging controls for input, make-up, and output. Logic Pro's Compressor is like this. I can drive it into the compressor hard with the input knob and have aggressive settings to squash the signal. Then, I can add makeup gain. But, this is before the limiter and distortion modules, and before the final output gain stage. Makeup gain and Output gain are two different gain stages.

UAD's "purity test" versions of the 1176 bluey has input and output, but no makeup. The classic SSL bus compressors have no input gain control (you would have to drive it with a separate gain utility), but have an output gain control that they call "makeup". Pro C 2 does not have input or make up, but do have an output. In reality, SSL bus comp makeup and Pro C 2's output are the same, and both are different from Logic Pro's make up control. Other compressors have bare-minimum controls. The One Knob plugins from Waves are like this. Just a drive (input) knob.

And, of course, some have an autogain feature. And some do that better than others.

Whatever you call the final gain stage (makeup, output, whatever), it's all just gain staging. And gain staging is just tedious, especially if you have a chain of plugins after the compressor. An autogain feature is a very welcome addition to handle this, when it's done well. Like, everyone on earth other than KVR aficionados feels this is a very useful option to have. You either have a good autogain function, or you deal with it manually. I don't know why anyone would enjoy gain staging manually rather than you know, push the Autogain button and move on with your mix. But maybe y'all have a perspective I have not considered.

Anyway... none of this changes the fact that Pro C has an Autogain feature, and it sucks, because it doesn't do what it says it does.

When you're talking about something that is explicitly designed to change the gain over the time it's not just simple gain staging.

Because it's time variant the energy distribution changes over time.

If you have a compressor fed with a drum loop with attack set to 30ms and release at 200 ms and then do the same but with 1ms attack and 20 ms release you can get that to average 6 db of gain reduction in both instances but they will differ in loudness because of how energy distributes.

An auto makeup gain in that instance will try to recover the 6 db often accounting for program dependency, knee etc. but in general the first will sound punchier but less loud and the latter will sound flatter but louder and denser.

If you add loudness levelling after that you have another measurement and reaction to account for which generally needs a longer window since our hearing integrates sound over time. Loudness is perceptually weighted integrated energy over time.

So systems that apply loudness leveling on top of compression will often lag behind the compressor’s time-variant gain changes and, in doing so, partially counteract the envelope shaping and transient control the compressor is actually producing.

Also saying that it's only "aficionados" that don't use auto gain is reductive. Pretty much every one of the most vaunted compressors in history do not use autogain.

Autogain is most useful where you have sources that have predictable dynamics and time constants are less critical and that's arguably where people might have (understandable) differences due to use case.

If you're using a compressor to shape dynamics, transients and feel then autogain/autolevelling can work against that whereas with dialogue and other more predictable signals autogain or auto levelling can be a time saver.

[masterhiggins]

A static implementation does what it says it does even if it's simply internally linked parameters based on assumed input strength.

Maybe your point isn't about the type of autogain used, but whether it works well, whatever method the developers go with. Personally I completely agree that I'd like improvement here, there's only upside to better autogain.

Maybe the best I've heard has been by PSP in Impressor. Very often, the default autogain value is spot-on and tracks the Compression knob very accurately. Actually, their implementation really outperforms FF/Presswerk/Cenozoix very noticeably (if memory serves - I've uninstalled Impressor).

Yup. It's based on their old Mixpressor, which I would still use every day if it didn't crap out on Sequoia. Excellent compressor. The perfect balance of flexibility and quickness.

[oobesan]

If you just want to compress the signal without changing the level, auto gain is supposed to do that for you. I mean, why would anyone consider that to be a bad thing?

I gotta say, if I'm understanding you correctly I think I would consider it a bad thing. Please forgive and correct me if I misunderstood, but what I think you are asking for is an auto-gain feature that would change the makeup gain dynamically based on the input signal. That in essence is additional compression, though perhaps quite slow, that some might rather refer to as "leveling". The problem with that is you would effectively have no control of this additional compression as it's just an on-off switch for a mysterious unknown algorithm with a preset configuration working behind the scenes. If your input signal is not consistent enough throughout the entire program material, you would likely be dealing with changes in level that are unpredictable at best and likely undesirable. Countering that by automating or somehow normalizing the input would be cumbersome and confusing, and might even require additional compensating automation on the output. The alternative of course is calculating a static makeup gain based on the current settings, with that calculation having been calibrated on some test signal deemed to be as representative as possible of basically every possible signal anyone is likely to feed the compressor. That's obviously impossible to get right as the exact same settings are likely to require different makeup gain for different signals. It's therefore far from perfect, but considering the alternative "dynamic" option it's also in my opinion the better choice. You'll usually still have to adjust the makeup gain anyway but it might at least get you closer to begin with. As bmanic suggested, that's probably what FabFilter (and likely many others as well) are doing. I will say, I have encountered a third option with a button that will let you sample some representative section of the signal for a few seconds and then set the makeup gain dial to a (static) position calculated by preforming some level-matching algorithm on that snippet with the current settings. Fircomp comes to mind but there may be others. Not sure what Newfangled are doing but I'm about to go check it out!

So, all I am talking about is gain staging. Different compressors handle this differently. Some compressors have separate gain staging controls for input, make-up, and output. Logic Pro's Compressor is like this. I can drive it into the compressor hard with the input knob and have aggressive settings to squash the signal. Then, I can add makeup gain. But, this is before the limiter and distortion modules, and before the final output gain stage. Makeup gain and Output gain are two different gain stages.

UAD's "purity test" versions of the 1176 bluey has input and output, but no makeup. The classic SSL bus compressors have no input gain control (you would have to drive it with a separate gain utility), but have an output gain control that they call "makeup". Pro C 2 does not have input or make up, but do have an output. In reality, SSL bus comp makeup and Pro C 2's output are the same, and both are different from Logic Pro's make up control. Other compressors have bare-minimum controls. The One Knob plugins from Waves are like this. Just a drive (input) knob.

And, of course, some have an autogain feature. And some do that better than others.

Whatever you call the final gain stage (makeup, output, whatever), it's all just gain staging. And gain staging is just tedious, especially if you have a chain of plugins after the compressor. An autogain feature is a very welcome addition to handle this, when it's done well. Like, everyone on earth other than KVR aficionados feels this is a very useful option to have. You either have a good autogain function, or you deal with it manually. I don't know why anyone would enjoy gain staging manually rather than you know, push the Autogain button and move on with your mix. But maybe y'all have a perspective I have not considered.

Anyway... none of this changes the fact that Pro C has an Autogain feature, and it sucks, because it doesn't do what it says it does.

When you're talking about something that is explicitly designed to change the gain over the time it's not just simple gain staging.

Because it's time variant the energy distribution changes over time.

If you have a compressor fed with a drum loop with attack set to 30ms and release at 200 ms and then do the same but with 1ms attack and 20 ms release you can get that to average 6 db of gain reduction in both instances but they will differ in loudness because of how energy distributes.

An auto makeup gain in that instance will try to recover the 6 db often accounting for program dependency, knee etc. but in general the first will sound punchier but less loud and the latter will sound flatter but louder and denser.

If you add loudness levelling after that you have another measurement and reaction to account for which generally needs a longer window since our hearing integrates sound over time. Loudness is perceptually weighted integrated energy over time.

So systems that apply loudness leveling on top of compression will often lag behind the compressor’s time-variant gain changes and, in doing so, partially counteract the envelope shaping and transient control the compressor is actually producing.

Also saying that it's only "aficionados" that don't use auto gain is reductive. Pretty much every one of the most vaunted compressors in history do not use autogain.

Autogain is most useful where you have sources that have predictable dynamics and time constants are less critical and that's arguably where people might have (understandable) differences due to use case.

If you're using a compressor to shape dynamics, transients and feel then autogain/autolevelling can work against that whereas with dialogue and other more predictable signals autogain or auto levelling can be a time saver.

I see what you're saying. Thanks for the thoughtful reply. I think I mentioned it before, but I'm not using compression to its fullest extent. I'm not recording live instruments or vocals (yet?). So for the most part, I may use it for a bit of transient control and to simply reduce dynamics on drums or plucky synths, or simply bus compression. I typically don't apply any make up gain in those use cases. The Pro C Autogain is just on as a default on INIT when you load it up. And it annoys me, because it's always about 2 dB louder that where I had the level set, and I have to take the extra steps of turning it off (now the signal is about 6-8 dB quieter) and adjust the output manually. The feature is a waste of time, basically.

[bmanic]

You know you can simply overwrite the default presets and have it start up exactly as you like it?

[NAD]

For reference, this is from the Pro-C manual (my emphasis):

Note that the auto-gain algorithm doesn't actually measure loudness to determine which gain to apply. It just makes an educated guess, so you might want still to tweak the output gain in some situations.

To be fair I feel like they're being pretty honest and not promising miracles.

For comparison I looked up some other manuals:

Auto Makeup: compensates for the drop in output level caused by gain reduction. It determines the theoretical compensation at a given threshold / ratio setting – for instance -20dB and 4:1 gives 5dB – and applies half of that value (2.5dB) to achieve the same perceived loudness.

Interestingly, U-He provides some insight into the actual calculation. (They know their audience: pedantic nerds )

AUTO: This button activates and deactivates the automatic make-up gain function of PSPImpressor. With this engaged, PSP Impressor will automatically adjust the output gain depending on the Shape, Compress and the AutoStrength trimpot on the hidden panel.

PSP seems to indicate explicitly which parameters affect the static calculation.

In order for the gain compensation engine to work best it must know the nominal level of your audio, which you can set with the NOM LVL control. You can simply turn this knob to set the level, or you can double click and type in a value. However, to make setting this level easiest we’ve added a LEARN button.

Newfangled looks to be doing that thing I mentioned that Fircomp does. Neat!

[Calagan]

I did read few posts criticizing the auto-gain in DMG TrackComp, but to me it's the only one I know that works great most of the times (sometimes it goes Berserk with transient heavy sources though). For exemple I don't like at all the TDR autogain that some here are praising - so it seems there's different conceptions of what autogain should be.

I'm not a coder, but to me it's not rocket science to take some LUFS or RMS Integrated value at the input, take the same value at the output and use the difference to set automatic make up gain. Doing this, you hear your source compressed more or less at the same level, so you can truly judge if it sounds better and not just louder.

This is actually what everybody is doing manually when trying to keep a consistant gain structure inside a plugin chain...
I'm surprised FabFilter can't manage something reliable...

[dionenoid]

I gotta say, if I'm understanding you correctly I think I would consider it a bad thing. Please forgive and correct me if I misunderstood, but what I think you are asking for is an auto-gain feature that would change the makeup gain dynamically based on the input signal. That in essence is additional compression, though perhaps quite slow, that some might rather refer to as "leveling". The problem with that is you would effectively have no control of this additional compression as it's just an on-off switch for a mysterious unknown algorithm with a preset configuration working behind the scenes. If your input signal is not consistent enough throughout the entire program material, you would likely be dealing with changes in level that are unpredictable at best and likely undesirable. Countering that by automating or somehow normalizing the input would be cumbersome and confusing, and might even require additional compensating automation on the output. The alternative of course is calculating a static makeup gain based on the current settings, with that calculation having been calibrated on some test signal deemed to be as representative as possible of basically every possible signal anyone is likely to feed the compressor. That's obviously impossible to get right as the exact same settings are likely to require different makeup gain for different signals. It's therefore far from perfect, but considering the alternative "dynamic" option it's also in my opinion the better choice. You'll usually still have to adjust the makeup gain anyway but it might at least get you closer to begin with. As bmanic suggested, that's probably what FabFilter (and likely many others as well) are doing. I will say, I have encountered a third option with a button that will let you sample some representative section of the signal for a few seconds and then set the makeup gain dial to a (static) position calculated by preforming some level-matching algorithm on that snippet with the current settings. Fircomp comes to mind but there may be others. Not sure what Newfangled are doing but I'm about to go check it out!

So, all I am talking about is gain staging. Different compressors handle this differently. Some compressors have separate gain staging controls for input, make-up, and output. Logic Pro's Compressor is like this. I can drive it into the compressor hard with the input knob and have aggressive settings to squash the signal. Then, I can add makeup gain. But, this is before the limiter and distortion modules, and before the final output gain stage. Makeup gain and Output gain are two different gain stages.

UAD's "purity test" versions of the 1176 bluey has input and output, but no makeup. The classic SSL bus compressors have no input gain control (you would have to drive it with a separate gain utility), but have an output gain control that they call "makeup". Pro C 2 does not have input or make up, but do have an output. In reality, SSL bus comp makeup and Pro C 2's output are the same, and both are different from Logic Pro's make up control. Other compressors have bare-minimum controls. The One Knob plugins from Waves are like this. Just a drive (input) knob.

And, of course, some have an autogain feature. And some do that better than others.

Whatever you call the final gain stage (makeup, output, whatever), it's all just gain staging. And gain staging is just tedious, especially if you have a chain of plugins after the compressor. An autogain feature is a very welcome addition to handle this, when it's done well. Like, everyone on earth other than KVR aficionados feels this is a very useful option to have. You either have a good autogain function, or you deal with it manually. I don't know why anyone would enjoy gain staging manually rather than you know, push the Autogain button and move on with your mix. But maybe y'all have a perspective I have not considered.

Anyway... none of this changes the fact that Pro C has an Autogain feature, and it sucks, because it doesn't do what it says it does.

When you're talking about something that is explicitly designed to change the gain over the time it's not just simple gain staging.

Because it's time variant the energy distribution changes over time.

If you have a compressor fed with a drum loop with attack set to 30ms and release at 200 ms and then do the same but with 1ms attack and 20 ms release you can get that to average 6 db of gain reduction in both instances but they will differ in loudness because of how energy distributes.

An auto makeup gain in that instance will try to recover the 6 db often accounting for program dependency, knee etc. but in general the first will sound punchier but less loud and the latter will sound flatter but louder and denser.

If you add loudness levelling after that you have another measurement and reaction to account for which generally needs a longer window since our hearing integrates sound over time. Loudness is perceptually weighted integrated energy over time.

So systems that apply loudness leveling on top of compression will often lag behind the compressor’s time-variant gain changes and, in doing so, partially counteract the envelope shaping and transient control the compressor is actually producing.

Also saying that it's only "aficionados" that don't use auto gain is reductive. Pretty much every one of the most vaunted compressors in history do not use autogain.

Autogain is most useful where you have sources that have predictable dynamics and time constants are less critical and that's arguably where people might have (understandable) differences due to use case.

If you're using a compressor to shape dynamics, transients and feel then autogain/autolevelling can work against that whereas with dialogue and other more predictable signals autogain or auto levelling can be a time saver.

Some of you guys are way overthinking things.

It doesn't matter whether it's precise or not, just getting in the ballpark will already be a huge time saver and enable much better judging of a signal.

Pro-C3's Autogain does not get anywhere near the ballpark. It causes huge volume differences immediately, and with every tweak, and/or mode switch. As reported by hundreds of ppl all over the net.

And this is not a typical Fabfilter issue, as it does work well enough in Pro-L2 and Pro-Q4. So i suspect that the method used in Pro-C3 wasn't tested well enough, and than especially not the addition of saturation.

But anyway, in my collection i have tools from DMG, TDR, Softube, Tone Projects, Melda, Acustica Audio, Acon Digital, Newfangled, and many more, which have Autogain that works for me. So i'll just keep using these tools, not in the least because i find Pro-C3 highly dissappointing in many areas anyway.

[E_Anderson]

I'm not a coder, but to me it's not rocket science to take some LUFS or RMS Integrated value at the input, take the same value at the output and use the difference to set automatic make up gain. Doing this, you hear your source compressed more or less at the same level, so you can truly judge if it sounds better and not just louder.

Most of this page of this thread is discussing the drawbacks of exactly that approach. But, being able to turn that type of autogain on, dial in your settings, and then turn it off, to lock in a static output gain & prevent those very drawbacks? Or have it time itself out after settling on a value? I agree that that method would improve a lot of plugins.

Interestingly, U-He provides some insight into the actual calculation. (They know their audience: pedantic nerds )

I wish I could be a smart nerd, instead

[Calagan]

I'm not a coder, but to me it's not rocket science to take some LUFS or RMS Integrated value at the input, take the same value at the output and use the difference to set automatic make up gain. Doing this, you hear your source compressed more or less at the same level, so you can truly judge if it sounds better and not just louder.

Most of this page of this thread is discussing the drawbacks of exactly that approach. But, being able to turn that type of autogain on, dial in your settings, and then turn it off, to lock in a static output gain & prevent those very drawbacks? Or have it time itself out after settling on a value? I agree that that method would improve a lot of plugins.

I don't know how it works for other plugins, but the point of the autogain in TrackComp is to compute the gain difference on some loop (or some significative part of your source) and then stop it so you've got your make-up gain setting. That stays there for the rest of the song (until you change it).
It doesn't make sense to have some kind of constant variation of the make up gain. I don't know how it works in Pro-C, but I can't believe FabFilter designed it to be used as a constantly moving target. It makes no sense at all.

[kraster]

Some of you guys are way overthinking things.

It doesn't matter whether it's precise or not, just getting in the ballpark will already be a huge time saver and enable much better judging of a signal.

Pro-C3's Autogain does not get anywhere near the ballpark. It causes huge volume differences immediately, and with every tweak, and/or mode switch. As reported by hundreds of ppl all over the net.

And this is not a typical Fabfilter issue, as it does work well enough in Pro-L2 and Pro-Q4. So i suspect that the method used in Pro-C3 wasn't tested well enough, and than especially not the addition of saturation.

But anyway, in my collection i have tools from DMG, TDR, Softube, Tone Projects, Melda, Acustica Audio, Acon Digital, Newfangled, and many more, which have Autogain that works for me. So i'll just keep using these tools, not in the least because i find Pro-C3 highly dissappointing in many areas anyway.

You know what else gets you in the ballpark?

Actually listening to the actual level in the context of a mix.

The idea that it's some kind of cardinal sin that a compressor adds or removes level is a relatively recent and quite bizarre phenomenon.

Not using a compressor because it doesn't have autogain pretty much rules out most of the best compressors ever made.

[MrJubbly]

Yeah, I think the Instance List will be expanded to other products. It seems very obvious to do that. The question is when they'll get around to it. Might be a bigger effort than it seems from the outside.

I believe the Instance List workflow feature would only make complete sense, once it eventually does become directly accessible from within each supported FabFilter plugin.

Since FabFilter have already added support within Pro-Q 4, for accessing Pro-C 3, Pro-G and Pro-DS, that may be a hint to indicate possible Instance List direct access within those plugins at least, in some future updates?

Whether full support will be added in future minor/major updates, only time will tell... But I can already imagine the improved workflow once all FabFilter products have been linked together via the Instance List. Should be groovy!

[cryophonik]

You know what else gets you in the ballpark?

Actually listening to the actual level in the context of a mix.

The idea that it's some kind of cardinal sin that a compressor adds or removes level is a relatively recent and quite bizarre phenomenon.

Not using a compressor because it doesn't have autogain pretty much rules out most of the best compressors ever made.

Thank you for saving me some typing!

Any time I see an "auto"-anything, by default, that means to me: click button, adjust to taste. It seems to me that too many people aren't aware of that second part.