The No.1 Website for Pro Audio
 Search This Thread  Search This Forum  Search Reviews  Search Gear Database  Search Gear for sale  Search Gearslutz Go Advanced
FabFilter Pro-L 2 experiences and tests Dynamics Plugins
Old 15th December 2017
  #1
Audio Alchemist
 
Lagerfeldt's Avatar
FabFilter Pro-L 2 experiences and tests

Hey guys

Just fooling around with the new FabFilter Pro-L 2.

Here are some of my findings and I'd love for other people to chime in with tests or answers (or questions) - especially if it's not just subjective.

Has the sound changed from Pro-L 1 to Pro-L 2?
Pro-L 2 will pass a null test with the old algorithms and settings, i.e. the Transparent, Punchy, Dynamic, and Allround styles, the Gain, Lookahead, Attack, Release, linking and Output ceiling parameters.

So no difference there, as in 100% identical.

The Oversampling parameter doesn't pass a null test, which is to be expected because according to the manual, oversampling was generally improved in the Pro-L 2, allowing it to catch more inter-sample peaks at the same rate.

The new styles and features will obviously sound different and provide new functionality not previously available in Pro-L.

True Peak Limiting
I'm not sure exactly how FabFilter are doing the TPL, but it's not just oversampling and it's not just an additional oversampled limiting stage - or at least it's not something that can easily be reproduced.

Activating manual oversampling in conjunction with TPL reduces the action of the TPL, that's confirmed.

Using TPL and setting an out ceiling in dBTP is more effective in respecting a True Peak value than activating any amount of manual oversampling (including 32x) and setting an out ceiling in dBFS, in my tests.

However, I need to make more tests to check exactly how well the TPL respects the TP value, but this is tricky to standardize since the only thing specified in ITU/EBU specs is to use a 4x oversampled meter.

Oversampling
Where as Pro-L 1 had 4x oversampling as the highest setting, Pro-L 2 now has 8x, 16x and 32x as well, and oversampling has been improved at all settings.

In my quick tests at 44.1 kHz I did not see an improvement going above 8x, i.e. 8x seemed to catch just as much as 16x and 32x in real life examples. Theoretically, ISPs can go much higher and if the CPU power is available I don't see any problem in going with the highest possible rate.

All oversampling takes place in the internal side-chain at the detection level only so the audio itself is not oversampled, just like in Pro-L 1.

Latency
Latency changes between some of the new styles which means that changing styles on the fly could cause the track to get out of sync until you press stop/start in the DAW. Doesn't matter much when mastering.

DC filter
The DC filter appears to be a regular static filter just above 0 Hz. Exactly where I don't know, but it doesn't sound like a steep filter.

I wouldn't use the DC filter unless there was a really good reason for it as it can easily cause the peak value of the input signal to raise by several dB, mainly if the source signal contains square waveforms (synths) or an already limited/clipped signal.

This would then result in much more limiting/pumping/distortion than otherwise.

-
More, still in progress.

Last edited by Lagerfeldt; 15th December 2017 at 09:23 AM..
Old 15th December 2017
  #2
Audio Alchemist
 
Lagerfeldt's Avatar
Figured I would take a look at the waveform when switching styles. Pretty unscientific, but here you go.

Considering how some of the descriptions of the styles in FabFilter's Pro-C2 compressor are completely wrong or misleading, I think this could be another justified case of "use your ears" when selecting a style.

All tests below use the exact same settings, unless otherwise noted. If memory serves me, it's 6 dB of gain reduction with 0.5 ms manual lookahead, 40 ms attack, 2 ms release, no linking of transient, full linking of release, no TPL, no oversampling.

Original test sine:



Transparent:



Punchy:



Allround with 0 attack (for comparison with Punchy above):



Dynamic:



Allround:



Aggressive - a word I would associate with near clipping rather than what both sounds like and looks like soft, lookahead limiting:



Modern:



Safe:



Bus:




Conclusion on the styles
Disclaimer: These oscilloscope shots don't capture any program-dependent limiting (if there's such a thing with the Pro-L 2), ISP handling, etc.

The new styles all appear to have additional lookahead, offsetting the manual lookahead parameter, which also explains the added latency. The new styles seemed to react very little to parameter changes compared to the old styles, which could indicate more "auto" going on behind the scenes or another factor that I'm not currently aware of. I'd like to know more about this.

Most of the styles probably work well for something. Personally I tend to opt for the Allround style and tweak that. Transparent seems to work well on some things, but in my experience it works best when the source isn't too compressed/limited/clipped already. Dynamic and Punchy seem to work well for more creative limiting.

The new Modern style sounds and measures very clean, but that's probably why it sounds best when it isn't pushed too much on transients since it'll quickly sound overcompressed, as a byproduct of keeping things so clean.

Last edited by Lagerfeldt; 15th December 2017 at 11:13 AM..
Old 15th December 2017
  #3
Interesting that you say it passes a null test with Pro-L 1 on the old styles - it doesn't here, at least in Punchy mode. All perimeters are identical, oversampling off.

Maybe I made a mistake somewhere...

EDIT: I just checked and there is definitely a difference, at least here, confirmed by a null test. I'd be interested in any others trying this for themselves to see if I'm correct, or if there is some kind of problem with my setup.

Settings on both Pro-L 1 and Pro-L 2:

Style: Punchy

Lookahead: 0.25ms

Attack: 1968.1 ms

Release: 271 ms

Channel Link Transients: 60%

Channel Link Release: 100%
Old 15th December 2017
  #4
Audio Alchemist
 
Lagerfeldt's Avatar
As a little side remark, all styles behave as expected when being fed a pre-clipped or square waveform, i.e. a square will still appear like a perfect square as long as TPL or oversampling aren't activated,

The Aggressive style will overcompress in a soft fashion, just like shown in the oscilloscope above, but it'll still be a perfect square.

However, as the only exception the Dynamic style does a weird little duck. It looks like an inverse Gibbs phenomenon , but not symmetrical.

Old 15th December 2017
  #5
Audio Alchemist
 
Lagerfeldt's Avatar
Quote:
Originally Posted by Hippocratic Mastering View Post
Interesting that you say it passes a null test with Pro-L 1 on the old styles - it doesn't here, at least in Punchy mode. All perimeters are identical, oversampling off.

Maybe I made a mistake somewhere...
100% null between Pro-L 1 and Pro-L 2 with all settings identical (no oversampling), including Punchy.

Just double checked now - 0 residue.

For good measure I tried extreme settings as well, but still there's a complete null between them.

FabFilter Pro-L 1 version 1.21 (64 bit)
FabFilter Pro-L 2 version 2.00 (64 bit)

AU version tested in Logic Pro X 10.3.3

Quote:
Originally Posted by Hippocratic Mastering View Post

Style: Punchy

Lookahead: 0.25ms

Attack: 1968.1 ms

Release: 271 ms

Channel Link Transients: 60%

Channel Link Release: 100%
100% null here.

Either you have a problem in your setup (PDC or similar, but I doubt that), an older version, you've missed something in the settings - or there's a bug/difference in one of the VST or AAX versions which doesn't exist in the AU versions.
Old 15th December 2017
  #6
Hmmmm. I haven't updated my installation of Pro-L 1 for a while - maybe they changed something in an update?

Are you null testing with tones or complex musical signals?
Old 15th December 2017
  #7
Audio Alchemist
 
Lagerfeldt's Avatar
Complex music signals, i.e. a full mix and I'm pushing it to get some action.
Old 15th December 2017
  #8
Hmmmm, very curious. I'll have to update to the latest version of Pro-L 1 and see if I get the same results, although I'm loathe to as there's a very specific thing that Pro-L 1 is doing for me in Punchy mode, at around 0.5db of gain reduction, that Pro-L 2 isn't quite as good at.
Old 15th December 2017
  #9
Audio Alchemist
 
Lagerfeldt's Avatar
VST on Mac also confirmed to null. Are you using AAX/ProTools?
Old 15th December 2017
  #10
I just tried null test, on Punchy they null to about -110dB, so basically completely. I'm on a Mac and use VST in Reaper. I like the modern setting and visual side of the Pro-L 2, but that got me thinking if it's worth the asking price.
Old 15th December 2017
  #11
Not using Pro Tools - VST on Reaper (Mac).
Old 15th December 2017
  #12
Audio Alchemist
 
Lagerfeldt's Avatar
Quote:
Originally Posted by cyjanopan View Post
I like the modern setting and visual side of the Pro-L 2, but that got me thinking if it's worth the asking price.
For me it's definitely worth the price, my upgrade price is only €47 from Pro-L.

I get some new styles to play around with and they might come in handy. The improved oversampling and especially True Peak Limiting and dBTP ceiling is great.

I'm not using the visuals at all in the Pro-L 2 and I have other ways of dealing with loudness measurements, but I'm sure that's a big plus for some people. That feature alone can cost several hundreds.

While I would have liked some more innovative styles it's hard to complain about a company that's doing good refinements to an already great product.
Old 16th December 2017
  #13
Lives for gear
 

Interesting to read all of this as I have the inside scoop on exactly how everything works.

I can say a few things though:

As before with Pro-L1, all algorithms are truly different from one another. Very different solutions, solving the same task (brickwall limiting).

All algorithms are highly program dependent (the new 'Modern' algorithm pushing this program dependency pretty much to the extreme in my opinion) so you won't be able to gather much from traditional methods of analyzing the output.

These are all still single-band limiting algorithms.

Oversampling is not there just for show or catching ISP. It makes a real difference in aliasing and in general behavior of the algorithm. You should be able to clearly hear the benefits if you compare non-oversampled to 32x oversampled signals that are pushing the limiter a bit. Listen to the mids and high-mids how they remain truer to the original at higher oversampling rates and don't get that added "fizz" that clogs up the depth imaging. This is where I think Pro L2 shines and is quite a step above the competition.

Then again I'm _extremely_ biased so keep that in mind. I'm also extremely sensitive and allergic to the mid/high-mid buildup that seems to happen quite easily with digital processors and thus has been trying to push the envelope in this area.

Some quick tips (and a sort of TL;DR):

- Compare non-oversampled to 32x oversampled critically. It's not there just to catch ISP.
- Experiment with "stupid" settings like 100% locked transient stereo linkage with 0% sustain linkage.
- The 'Modern' algorithms look-a-head is quite different to the rest of the algorithms. Don't be afraid of pushing it all the way up to max while instead opening up the attack a lot (don't look at the values.. think of it as a knob with full range of usefulness).
- Explore the presets, especially in the 'Basic' folder. They show common usage scenarios for the algorithms.

Cheers!
Old 16th December 2017
  #14
Gear Guru
 
Brent Hahn's Avatar
 

Quote:
Originally Posted by bmanic View Post
Interesting to read all of this as I have the inside scoop on exactly how everything works.
Since you're here, maybe a dumb question. "Bypass" in Pro-L1 doesn't get rid of latency. Assuming that's not operator error, is this fixed in the new version?
Old 16th December 2017
  #15
Lives for gear
 

I'll definitely confirm a sonic difference/improvement with Oversampling from 4x, 8x, 16x, out to 32x.

Blatantly obvious. Worth the upgrade alone. Also, really liking the new Metering.
Old 16th December 2017
  #16
Lives for gear
 
da goose's Avatar
Quote:
Originally Posted by bmanic View Post
Interesting to read all of this as I have the inside scoop on exactly how everything works.

I can say a few things though:

As before with Pro-L1, all algorithms are truly different from one another. Very different solutions, solving the same task (brickwall limiting).

All algorithms are highly program dependent (the new 'Modern' algorithm pushing this program dependency pretty much to the extreme in my opinion) so you won't be able to gather much from traditional methods of analyzing the output.

These are all still single-band limiting algorithms.

Oversampling is not there just for show or catching ISP. It makes a real difference in aliasing and in general behavior of the algorithm. You should be able to clearly hear the benefits if you compare non-oversampled to 32x oversampled signals that are pushing the limiter a bit. Listen to the mids and high-mids how they remain truer to the original at higher oversampling rates and don't get that added "fizz" that clogs up the depth imaging. This is where I think Pro L2 shines and is quite a step above the competition.

Then again I'm _extremely_ biased so keep that in mind. I'm also extremely sensitive and allergic to the mid/high-mid buildup that seems to happen quite easily with digital processors and thus has been trying to push the envelope in this area.

Some quick tips (and a sort of TL;DR):

- Compare non-oversampled to 32x oversampled critically. It's not there just to catch ISP.
- Experiment with "stupid" settings like 100% locked transient stereo linkage with 0% sustain linkage.
- The 'Modern' algorithms look-a-head is quite different to the rest of the algorithms. Don't be afraid of pushing it all the way up to max while instead opening up the attack a lot (don't look at the values.. think of it as a knob with full range of usefulness).
- Explore the presets, especially in the 'Basic' folder. They show common usage scenarios for the algorithms.

Cheers!
That! But I must confess that i'm also a bit biased.
Old 16th December 2017
  #17
Lives for gear
 

Quote:
Originally Posted by Brent Hahn View Post
Since you're here, maybe a dumb question. "Bypass" in Pro-L1 doesn't get rid of latency. Assuming that's not operator error, is this fixed in the new version?
I'm sure that is on purpose. The reason being that if the built in bypass button in Pro-L would get rid of the latency then you would have a horrible "click" while the DAW reassigns the latency compensation. So having the plugin at the same latency when in internal bypass is actually a smart move.. then it can be safely automated.

Use the DAW built in bypass if you need to get rid of latency. Some DAWs don't allow dynamic latency compensation during playback though.
Old 16th December 2017
  #18
Audio Alchemist
 
Lagerfeldt's Avatar
Quote:
Originally Posted by bmanic View Post
Interesting to read all of this as I have the inside scoop on exactly how everything works.
You tease, spill the beans on the TPL or GTFO ;-)

Quote:
[...]the new 'Modern' algorithm pushing this program dependency pretty much to the extreme in my opinion) so you won't be able to gather much from traditional methods of analyzing the output.
I think that's very true. Definitely a case of "use your ears".

Quote:
Oversampling is not there just for show or catching ISP. It makes a real difference in aliasing and in general behavior of the algorithm. You should be able to clearly hear the benefits if you compare non-oversampled to 32x oversampled signals that are pushing the limiter a bit. Listen to the mids and high-mids how they remain truer to the original at higher oversampling rates and don't get that added "fizz" that clogs up the depth imaging. This is where I think Pro L2 shines and is quite a step above the competition.
That's also true, oversampling will reduce aliasing and distortion. However, oversampling comes with a downside as well, so it's a trade-off, but fortunately the loudness war is slowly coming to an end.

I think this focus on TPL and oversampling (and loudness metering) shows that FF knows what's right around the corner.
Old 16th December 2017
  #19
Gear Guru
 
Brent Hahn's Avatar
 

Quote:
Originally Posted by bmanic View Post
I'm sure that is on purpose. The reason being that if the built in bypass button in Pro-L would get rid of the latency then you would have a horrible "click" while the DAW reassigns the latency compensation.
Okay. But if the latency is always there and I'm in mid-mix and the client decides they want to change or add a part, I have to...

1. Create a new, second master channel, minus Pro-L.
2. Assign the new master channel to control 2 unused buses.
3. Reassign the "real" master channel to control 2 other unused buses.
4. Reassign the new master channel to control the original mix buses.
5. Jack up the master fader so the monitor and cue levels are healthy.
6. Record the overdubs.
7. Undo 5, 4, 3 and 2 in that order.

Quote:
... having the plugin at the same latency when in internal bypass is actually a smart move.. then it can be safely automated.
Who, in real life, is going to automate the bypass toggle? I suppose if you wanted to suddenly change Pro-L settings in mid-mix, you could run two instances on the same mix bus and use bypass to switch back and forth, but that would be equally difficult and crazy. Far better to write automation top to bottom with the initlal setting, and then do a second write with your alternate settings to your selected range(s) on the timeline. That's how I do it, anyway.
Old 16th December 2017
  #20
Audio Alchemist
 
Lagerfeldt's Avatar
Pretty much the standard way of doing it, i.e. the internal bypass button (as opposed to the DAW GUI one) keeps the latency fixed. I prefer it like that.
Old 16th December 2017
  #21
Lives for gear
 

Quote:
Originally Posted by Brent Hahn View Post
Okay. But if the latency is always there and I'm in mid-mix and the client decides they want to change or add a part, I have to...

1. Create a new, second master channel, minus Pro-L.
2. Assign the new master channel to control 2 unused buses.
3. Reassign the "real" master channel to control 2 other unused buses.
4. Reassign the new master channel to control the original mix buses.
5. Jack up the master fader so the monitor and cue levels are healthy.
6. Record the overdubs.
7. Undo 5, 4, 3 and 2 in that order.



Who, in real life, is going to automate the bypass toggle? I suppose if you wanted to suddenly change Pro-L settings in mid-mix, you could run two instances on the same mix bus and use bypass to switch back and forth, but that would be equally difficult and crazy. Far better to write automation top to bottom with the initlal setting, and then do a second write with your alternate settings to your selected range(s) on the timeline. That's how I do it, anyway.
I still don't understand the problem.. you have two ways of doing it, you can bypass the plugin from your DAW, which would then of course cancel the latency.. or internally within Pro-L (and keep the latency). This way you have BOTH choices. If Pro-L cancelled the latency there'd be no choice.
Old 16th December 2017
  #22
Gear Guru
 
Brent Hahn's Avatar
 

Quote:
Originally Posted by bmanic View Post
I still don't understand the problem.. you have two ways of doing it, you can bypass the plugin from your DAW, which would then of course cancel the latency.. or internally within Pro-L (and keep the latency). This way you have BOTH choices. If Pro-L cancelled the latency there'd be no choice.
I heard you the first time. I was waiting to respond until I had tried it on my system. I didn't know the two kinds of bypass worked differently.
Old 16th December 2017
  #23
Motown legend
 
Bob Olhsson's Avatar
 

Verified Member
They don't work differently in Pro Tools unless you disable the plug-in.
Old 16th December 2017
  #24
Lives for gear
 

Yeah it depends on the DAW. In some you can configure how it should act. Most reset the latency compensation upon stop/start of playback. Some are brutal and "glitch it" as soon as you turn off the plugin, even during playback.
Old 17th December 2017
  #25
Gear Guru
 
Thomas W. Bethe's Avatar
 

Verified Member
Thanks for all the scientific work. I really like the way it sounds and the new interface is very well done. I think they have a winner on their hands.

Again thanks for the "scientific" approach to what it is doing to the sound. It confirms in a good way what I am hearing.
Old 13th March 2018
  #26
Gear Maniac
Comparison

Thank you to
Lagerfeldt for the diligent and thorough research

Any thoughts from those who have been using Pro-L2 on sonic and functional comparisons to other contenders (e.g. Limitless, Elevate, Xenon, ISL2, Ozone, etc.)? I've been using Xenon for years, but am looking into other possible mastering processors, especially those (like Pro-L2) that provide and informative graphical feedback, a "difference" listening mode, etc.
Topic:
Post Reply

Welcome to the Gearslutz 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…
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Forum Jump
Forum Jump