Presswerk Manual

[Local Man]

I am reading the presswerk manual and it is going so far over my head. I thought I knew some things!

Some bewildering moments for me so far:

-The compression ratio in feed back mode is limited to about 2.5:1? (from the manual: 'we chose not to allow more than about 2.5:1 ratio in FB mode') Is that the actual compression ratio because clearly the knob still goes to 20:1?

-These sentences in regard to feed back mode make almost zero sense to me:
'To limit the input signal, the feedback loop would need masses of extra gain. In fact the gain should be infinite, and we all know that working with infinity can be “rather problematic”!
Some designs offer a huge amount of gain anyway, which is released as soon as saturation happens.'

Again I thought I knew some things about compression but these sentences are like a foreign language to me.

-This bit about M/S mode:
'Dynamics processing in M/S mode is independent of the selected EDIT mode. Even with Link, the channels will still operate independently unless you set Channel Link to maximum.'
When the edit mode is in link, the mid and side channels are being processed in exactly the same way so the result should be no different than stereo, right? At least that is what I am experiencing when using it in link mode. It is only when I turn link off that I can hear an independant mid and side signal changing as I change setting in either channel 1 or channel 2.
In other words, when link is enabled, changing settings in either channel 1 or channel 2 results in the exact same change to the signal.
But the above quoted sentence makes it sound as if that shouldn't be true. It says the channels will 'still operate independently' but they don't.

I don't know, maybe I just need to get some sleep so that's where I've stopped for now.
I have to say; I know very well how compressors work (practically speaking) and normally when I go through the manual for an audio processor of any sort there is not that much information that is new to me.
But here I feel like I am being bombarded with things that I am clueless about.
Maybe I really need to seek out some further education the subject.

[lunardigs]

Good questions! I'm interested to hear the answers to this also. I use Presswerk on nearly every project, M/S processing as well.
Wish I could elucidate.

[Urs]

Unfortunately Sascha is on vacation for another week or two and I'm not sure I know enough about PWs inner workings to explain this any better :-/

[Local Man]

OK, well now I don't feel quite so bad about my lack of understanding.
Thanks for you response.

I've continued to go through the manual from there and while much of it makes sense, it still feels like I am watching a cricket match yet I only know the rules of baseball.

Either way Presswerk is a fantastic compressor. I love the sound and it is so deep. And despite the confusion regarding the under the hood details, the compressor itself is fairly intuitive once you spend some time with it. I was a bit intimidated at first and relying only on the presets but now I am starting to realize it's full potential.

[dermage]

Link mode means the controls from both channels are linked, not the compression detector, as far as I understand it. That means the audio is still acting as summed until the channel link is in 0%.
Channel link is the compression behavior. That's why it's stated, that the one has nothing to do with the other (GUI control linking).

Btw: Just read the complete "Channel Link" and "Edit" chapters again. I find it quite clear how this is differentiated.

[Local Man]

Link mode means the controls from both channels are linked, not the compression detector, as far as I understand it. That means the audio is still acting as summed until the channel link is in 0%.
Channel link is the compression behavior. That's why it's stated, that the one has nothing to do with the other (GUI control linking).

Btw: Just read the complete "Channel Link" and "Edit" chapters again. I find it quite clear how this is differentiated.

Yes, there is no confusion for me there. I understand the difference between Link mode (controls move together) and channel link (channels detected seperately and thus compressed seperately).
What I was confused about was the statement 'Dynamics processing in M/S mode is independent of the selected EDIT mode,' where I take that to mean link mode doesn't apply. But it clearly does.
I could just be misunderstanding that part though. In the next sentence, channel link is mentioned and maybe that is what is being referred to the whole time. I'll have to mess with it again to see if that makes sense.

Yes, it is confusing having "link" mean two different things when talking about it in words but in practice it is not that confusing as they are clearly two different things.

[Teawhyelleare]

Decided to start a read of the manual after reading this thread. I recently obtained presswerk, and even though I've been using it like it may go out of fashion (it won't!), I hadn't looked through the manual. I am already finding value in this decision, so thanks for the nudge

So far, I want to give s'more kudos to u-he for creating such an amazing plugin and for providing such solid support. At first read, I find the manual very well written, useful, and entertaining. All of the lines that were quoted here make perfect sense to me in context... so I started typing out a response to @Local Man, assuming that the misunderstanding was in what is written, and I thought I would jump in to help... but then I opened up presswerk on an audio channel to play along while I did my best to articulate my thoughts, an lo and behold, I am now just as confused as you, in the same way that you described. The words make perfect sense on their own, and presswerk seems to make sense on its own, however the two do not seem to be in agreement with one another...

Adjusting the ratio knob in different detection modes does not provide the behavior that the text suggests.

&

I totally get what you mean with the Channel Link disparity. Again, the text makes sense, presswerk makes sense, but they are "saying" two totally different things. Unless I am missing the same things that you are?

[Urs]

Sascha is back tomorrow or so, maybe he can shed some light on this

[sascha]

On ratio with FB mode:
The amount of gain reduction (-> ratio) in feed-back mode is indeed not identical to the one applied in feed-forward detection. Mainly, this is because in FF mode, the detection process can operate on the incoming signal and cover its entire dynamic range. In FB mode, all the detector 'sees' is an already reduced dynamic range, so there's less of 'work' to do. This was a good thing for nonlinear voltage-regulation devices because linear operation usually is within a limited range in, say, a field-effect transistor.
So, what Presswerk does here is simply exchange the signal at its detector input, with no further pre-processing dependent on the origin. This is a bit different to hardware devices that strictly operate in either mode. When you have a FB compressor you can try shaping the detector's control signal exponentially so that it forces the gain-reduction element to do even more work with more GR, resulting in higher ratios. Although this might interfere with a GR element's nonlinearity, and - for instance - you can observe this on an 1176 when it backs off again past a certain input level.

(This is all perhaps difficult to explain, but let me try with an analogy: your home room thermostat. Imagine you're freezing in winter time, cranking up the knob on the wall regulator. As a result, the room heats up, but it might still be too cold because it's so freezing outside... Imagine you putting the thermostat's sensor on the outside wall; now the room will heat up faster, and it likely might heat up more. But also, you'd encounter more fluctuation as the outside temperature changes during the day (sun/shadow/wind etc.). If you set the regulator to 20°C, it will surely achieve it using outside measurement. But once put inside, it will never reach it exactl and always be a bit lower. Inside, it's a feed-back system.)

Of course, we could compensate for this 'auto ride' in FB mode automatically in software, to some extent, and working in the LOG domain makes it even more simple. But compensating doesn't change the fact that the envelope is applied twice, and thus the overall sound is so different. The same ratio on a FB & FF compressor results in completely different perception of signal envelope, a FB device then tends to 'suck' out the transients, kind of 'hyperactive'. While this is sometimes desirable and a key aspect to devices like an 1176, it is difficult with a general-purpose design like in Presswerk. So we chose to not have it here.

You can push things a bit on PW, though: apply some negative gain reduction via the NonLin control, and this adds a bit of additional ratio while retaining predictive results in terms of envelope contour.

On the LINK mode:
When you have the button engaged and therefore both channels processing the M & S part equally, it doesn't mean the audio is 'linked'. It's only the controls' parameters that are matched. If you feed PW a stereo source with highly-correlated material, like 0-45° or so, it won't matter that much how far the parameters are set apart. But the more uncorrelated the M & S partials are, the more an 'unlinked' parameter set matters. This is where the Channel Link knob comes in: turn it to 100% and you're left with only one detection signal. Then it doesn't matter if the controls are linked or not. But when you have it on 0% it matters the most whenever a great deal of correlation is present in the incoming source.

I hope I could clear this up a bit more. If not, say so

[Teawhyelleare]

OK, well now I don't feel quite so bad about my lack of understanding.
Thanks for you response.

I've continued to go through the manual from there and while much of it makes sense, it still feels like I am watching a cricket match yet I only know the rules of baseball.

Either way Presswerk is a fantastic compressor. I love the sound and it is so deep. And despite the confusion regarding the under the hood details, the compressor itself is fairly intuitive once you spend some time with it. I was a bit intimidated at first and relying only on the presets but now I am starting to realize it's full potential.

I've been patiently waiting for the OP to reply as to not hijack the thread, but since there is nothing but silence here, and you u-he dudes generously took the time to reply in such a timely and verbose manner, I'll give it a whirl. Responses incoming...

[Local Man]

OK, well now I don't feel quite so bad about my lack of understanding.
Thanks for you response.

I've continued to go through the manual from there and while much of it makes sense, it still feels like I am watching a cricket match yet I only know the rules of baseball.

Either way Presswerk is a fantastic compressor. I love the sound and it is so deep. And despite the confusion regarding the under the hood details, the compressor itself is fairly intuitive once you spend some time with it. I was a bit intimidated at first and relying only on the presets but now I am starting to realize it's full potential.

I've been patiently waiting for the OP to reply as to not hijack the thread, but since there is nothing but silence here, and you u-he dudes generously took the time to reply in such a timely and verbose manner, I'll give it a whirl. Responses incoming...

Thank you for reminding me of this thread (I forgot about it way down here ).

Sascha, thank you very much for your clear and detailed explanation. I still need to sit down in front of Presswerk as I reference your post to make sure I fully understand but just based on what you wrote, I understand in a way that I didn't before. It is clear though that there is still much that I need to learn but your post makes it clear for me where my gaps of knowledge lie (that is I now know what I don't know a little better ).

[Teawhyelleare]

On ratio with FB mode:
The amount of gain reduction (-> ratio) in feed-back mode is indeed not identical to the one applied in feed-forward detection. Mainly, this is because in FF mode, the detection process can operate on the incoming signal and cover its entire dynamic range. In FB mode, all the detector 'sees' is an already reduced dynamic range, so there's less of 'work' to do. This was a good thing for nonlinear voltage-regulation devices because linear operation usually is within a limited range in, say, a field-effect transistor.
So, what Presswerk does here is simply exchange the signal at its detector input, with no further pre-processing dependent on the origin. This is a bit different to hardware devices that strictly operate in either mode. When you have a FB compressor you can try shaping the detector's control signal exponentially so that it forces the gain-reduction element to do even more work with more GR, resulting in higher ratios. Although this might interfere with a GR element's nonlinearity, and - for instance - you can observe this on an 1176 when it backs off again past a certain input level.

(This is all perhaps difficult to explain, but let me try with an analogy: your home room thermostat. Imagine you're freezing in winter time, cranking up the knob on the wall regulator. As a result, the room heats up, but it might still be too cold because it's so freezing outside... Imagine you putting the thermostat's sensor on the outside wall; now the room will heat up faster, and it likely might heat up more. But also, you'd encounter more fluctuation as the outside temperature changes during the day (sun/shadow/wind etc.). If you set the regulator to 20°C, it will surely achieve it using outside measurement. But once put inside, it will never reach it exactl and always be a bit lower. Inside, it's a feed-back system.)

Of course, we could compensate for this 'auto ride' in FB mode automatically in software, to some extent, and working in the LOG domain makes it even more simple. But compensating doesn't change the fact that the envelope is applied twice, and thus the overall sound is so different. The same ratio on a FB & FF compressor results in completely different perception of signal envelope, a FB device then tends to 'suck' out the transients, kind of 'hyperactive'. While this is sometimes desirable and a key aspect to devices like an 1176, it is difficult with a general-purpose design like in Presswerk. So we chose to not have it here.

You can push things a bit on PW, though: apply some negative gain reduction via the NonLin control, and this adds a bit of additional ratio while retaining predictive results in terms of envelope contour.

On the LINK mode:
When you have the button engaged and therefore both channels processing the M & S part equally, it doesn't mean the audio is 'linked'. It's only the controls' parameters that are matched. If you feed PW a stereo source with highly-correlated material, like 0-45° or so, it won't matter that much how far the parameters are set apart. But the more uncorrelated the M & S partials are, the more an 'unlinked' parameter set matters. This is where the Channel Link knob comes in: turn it to 100% and you're left with only one detection signal. Then it doesn't matter if the controls are linked or not. But when you have it on 0% it matters the most whenever a great deal of correlation is present in the incoming source.

I hope I could clear this up a bit more. If not, say so

Thank you for the detailed and fascinating reply, Sascha, and thanks for being on top of everything as always, Urs. It's really cool of you to respond so quickly and so thoroughly Sending you guys heaps love and gratitude, as usual.

Sascha, I have uploaded a quick video for demonstration. What you wrote makes some sense, especially from a theory standpoint, and while I don't doubt that there is wonderful wizardry under the hood, I'm still not following.

https://vimeo.com/349236017

Basically, the detection method seems to have 0 effect on presswerk's behavior on most material, at most any setting. I believe that what OP was trying to express is that if feedback mode is processing differently, and is restricted to a 2:1ish ratio, then it is odd that the knob still goes to 20, and the reduction meters (and indeed the measured output) are nearly identical between modes. This confuses me as well. The manual paints FB and INT as completely different animals from FF, but presswerk's behavior does not seem to reflect this.

I like reading what you wrote, but frankly, it hasn't cleared up my confusion or helped me to understand presswerk's behavior in this instance better. Granted, I believe I can notice subtle differences between the modes on different material, I notice the "hyperactive suck" on drums, for example, your description of the different characters is quite helpful, and your general compressor knowledge is enlightening. What you are saying in relation to presswerk just doesn't make sense to me yet.

I'm going to leave the link buttons discussion to you lot for now, as I think that I grasp this. I'm still not sure that the manual language and the behavior reflect one another, but I'm going to have to focus on it for a bit, and I don't have the time to do it at the moment

Thanks again! Please let me know what I am missing here.. In simple terms, I would be most interested in what presswerk is actually doing here. Learning about design implementations and decisions made against other designs is super interesting, and I'm all about digging in to it, but I would like to establish a baseline understanding of PWs behavior as it's own thing first. Trying to understand nuanced perceptual discrepancies in presswerk's basic features is challenging enough without introducing multiple other concepts and compressors, for me at least

[Teawhyelleare]

OK, well now I don't feel quite so bad about my lack of understanding.
Thanks for you response.

I've continued to go through the manual from there and while much of it makes sense, it still feels like I am watching a cricket match yet I only know the rules of baseball.

Either way Presswerk is a fantastic compressor. I love the sound and it is so deep. And despite the confusion regarding the under the hood details, the compressor itself is fairly intuitive once you spend some time with it. I was a bit intimidated at first and relying only on the presets but now I am starting to realize it's full potential.

I've been patiently waiting for the OP to reply as to not hijack the thread, but since there is nothing but silence here, and you u-he dudes generously took the time to reply in such a timely and verbose manner, I'll give it a whirl. Responses incoming...

Thank you for reminding me of this thread (I forgot about it way down here ).

Sascha, thank you very much for your clear and detailed explanation. I still need to sit down in front of Presswerk as I reference your post to make sure I fully understand but just based on what you wrote, I understand in a way that I didn't before. It is clear though that there is still much that I need to learn but your post makes it clear for me where my gaps of knowledge lie (that is I now know what I don't know a little better ).

Hah! No worries. Welcome back

What a nice reply

Does anyone know if there is a way to tag users without quoting their whole post? Seems rather clunky to have to quote someone's whole post to get their attention, and I've noticed it's easy to lose track of stuff around here. (I've also abandoned some (rather in depth) posts I've tried to make after the forum eats them multiple times, but that's a chat for a different time and place.)

[yellowmix]

Cut out any text but leave the opening and closing quote blocks.

[Teawhyelleare]

That's clever Thanks!