Post-fader plugins inserts

[mschnell]

Originally Posted by Gass n Klang

I think the easiest way (handlingwise) would be a "fader plugin slot" in the insert area. The fader slot should be the last by default. So every added plugin would sit before (and therefor is pre-fader by default). But we could move it post fader manually. Dead easy to handle but I have no idea how complicated it is to implement.

Yep this is THE "Reapery" way of life.
(I do suppose ther will be some issues using automation etc, but this should be solvable.)

[Trakz]

Since post fader never happened. Can fx container be used post fader ?

[mschnell]

No

[MrMahn]

+1, would be extremely helpful for airwindows console.

[goateeproductions]

+1

[dna598]

Want/need.

+1

[bountyshores]

+1 for this!!!

It will make REAPER absolutely perfect for my workflow (Softube Console 1 + Airwindows Console).

[Junkyard Sam]

I hope it's not inappropriate to bump this, but I need Post-fader plugin inserts desperately.

In the meanwhile, I have an alternative solution:

Enable "Show embedded UI in MCP" in ZenoMOD VU meter and you can drag on the VU meter just like a volume knob!

Do that, and then insert the ZenoMOD VU meter BEFORE your "post fader fx" and use the VU Meter as your 'fader' instead of Reaper's fader.

I made some tweaks to the ZenoMOD VU meter that make it more useful for this (-96 to +18db instead of the default -18 to +18, and I changed some drag behavior) --- but this approximates the post-fader fx experience in a non-cumbersome way.

To be clear, you just drag up and down on the VU meter to change volume prior to the 'post fader fx' such as driving into a compressor or console emulation.

But still, we NEED Post-Fader FX!!! And there's nothing to lose... People who don't want or need them don't have to use it.

Proof it's needed, though, is the popularity of console emulation effects. You know how AirWindows Console8 has a "don't use your DAW's faders" workflow? Truth is, almost all console emulation plugins work better that way. Most developers won't tell you, though, because it's a bad workflow if your DAW doesn't support post-fader FX.

So who needs this? ANYONE that likes mixing into a compressor or console emulation plugins. That's a lot of people, and no --- adding an option for post fader effects doesn't make Reaper "overly complex."

PS. I might be able to share my ZenoMOD VU Meter modifications for anyone that wants them, although I'm not sure about etiquette with such a thing so maybe not.

[goateeproductions]

+1

[mozart999uk]

+1

[goateeproductions]

I wrote a script and plugins for fake post-fader insert functionality using just one channel. It's, of course, not a true post-fader insert, but it does the job in most cases.

Repo here:
https://github.com/AudioIdeas/AudioI...ster/index.xml

Install and overview video here:
https://www.youtube.com/watch?v=Lvo_vt5nG9I

[MonkeyBars]

Originally Posted by goateeproductions

I wrote a script and plugins for fake post-fader insert functionality using just one channel. It's, of course, not a true post-fader insert, but it does the job in most cases.

Repo here:
https://github.com/AudioIdeas/AudioI...ster/index.xml

Install and overview video here:
https://www.youtube.com/watch?v=Lvo_vt5nG9I

This is great, well done! Going the JSFX route is very clever. I had a much more complex solution wtihout custom JSFX utilizing ReaLearn and a tone generator etc etc but yours is much more elegant. The helper is another great addition – even ReaLearn can't use Reaper's track volume as a control input. Really glad I don't need to do my solution now!

Edit: Ah – I see you're actually not utilizing channels here to keep the signal at unity after the pseudo-post-fader FX. In fact, my (unreleased) solution (which is pretty janky and would be considerably smoother if helgoboss implemented source as target in ReaLearn) does route the audio properly so the user gets a true post-fader FX emulation in audio. But considering this, why did you mention AirWindows Console plugins? They specifically require unity to be maintained between the Out and In plugins. If your Post-fader-insert-end plugin is inverting the gain effect of the -start plugin, unity won't be maintained and an AW Console plugin won't produce the desired sound.

[bountyshores]

Originally Posted by goateeproductions

I wrote a script and plugins for fake post-fader insert functionality using just one channel. It's, of course, not a true post-fader insert, but it does the job in most cases.

Repo here:
https://github.com/AudioIdeas/AudioI...ster/index.xml

Install and overview video here:
https://www.youtube.com/watch?v=Lvo_vt5nG9I

Originally Posted by MonkeyBars

But considering this, why did you mention AirWindows Console plugins? They specifically require unity to be maintained between the Out and In plugins. If your Post-fader-insert-end plugin is inverting the gain effect of the -start plugin, unity won't be maintained and an AW Console plugin won't produce the desired sound.

I don't know if it's mathematically perfect (probably not) like Airwindows PurestGain or BitShiftGain, but it seems to be unity.

For example, Post-fader-insert-start (-6dB) -> AWConsoleChannel -> Post-fader-insert-end (+6dB) -> DAW Fader (-6dB) -> AWConsoleBuss.

Am I missing something?

[MonkeyBars]

Originally Posted by bountyshores

I don't know if it's mathematically perfect (probably not) like Airwindows PurestGain or BitShiftGain, but it seems to be unity.

For example, Post-fader-insert-start (-6dB) -> AWConsoleChannel -> Post-fader-insert-end (+6dB) -> DAW Fader (-6dB) -> AWConsoleBuss.

Am I missing something?

you lose 6dB of resolution in the signal. definitely lossy

[bountyshores]

Originally Posted by MonkeyBars

you lose 6dB of resolution in the signal. definitely lossy

Well, I guess it's beyond my comprehension. I wonder it's like it won't work completely or just not as perfectly.

[MonkeyBars]

Originally Posted by bountyshores

I wonder it's like it won't work completely or just not as perfectly.

the latter – all depends on your standards for exactness.

decreasing a signal's gain by 6dB, applying an effect, then giving it 6dB of gain back results in a loss of those bits of volume.

For 16bit it's a serious degradation, while for 32bit it might be all but inaudible.

[ScuzzyEye]

Originally Posted by MonkeyBars

the latter – all depends on your standards for exactness.

decreasing a signal's gain by 6dB, applying an effect, then giving it 6dB of gain back results in a loss of those bits of volume.

For 16bit it's a serious degradation, while for 32bit it might be all but inaudible.

In 32-bit floating point subtracting 6.02 dB and then adding back 6.02 dB is the same as /2 then *2. That will be lossless. Other non multiples of 6 dB might have small rounding errors, but that'll result in distortion down around -144 dB. It won't be a whole bit lost like it would in integer.

[bountyshores]

Oh I got it now. Thanks for the explanation, MonkeyBars and ScuzzyEye!

[MonkeyBars]

Originally Posted by ScuzzyEye

In 32-bit floating point subtracting 6.02 dB and then adding back 6.02 dB is the same as /2 then *2. That will be lossless.

Agreed, but that's not the case here. We're speaking about subtracting 6.02 dB then applying an effect then adding back 6.02 dB. That is lossy.

[ScuzzyEye]

Originally Posted by MonkeyBars

Agreed, but that's not the case here. We're speaking about subtracting 6.02 dB then applying an effect then adding back 6.02 dB. That is lossy.

As long as the effect is linear (not a compressor or distortion with some sort of threshold) plug-ins get passed floating point values like everything else in Reaper. You can change level by orders of magnitude (48 dB) up and down and not lose a bit of precision.

I just did an experiment. Sent pink noise to two tracks (and turned off parent send on the source), and inverted the phase of one of them. Added an EQ to one, changed some values with a boost and a cut, and then duplicated that EQ on the other. Now on one track I added a Volume Adjustment, and set it to +6 before the EQ, and then another after the EQ with -6 dB. They null perfectly (-inf). Actually the Volume Adjustment JS effect nulls perfectly with itself, but the track fader set to -6.00 dB doesn't perform the same math as the effect when set to the same value, and nulls to around -70 dB.

So it's probably better to use the same process to add and subtract level, but -70 dB isn't a huge worry.

[MonkeyBars]

Originally Posted by ScuzzyEye

As long as the effect is linear (not a compressor or distortion with some sort of threshold)

We are speaking here of AirWindows's ConsoleXYZ plugins, which are definitively nonlinear – it's kind of the whole point of using them.

the Volume Adjustment JS effect nulls perfectly with itself, but the track fader set to -6.00 dB doesn't perform the same math as the effect when set to the same value, and nulls to around -70 dB.

So it's probably better to use the same process to add and subtract level, but -70 dB isn't a huge worry.

Go ahead and download Console8Channel and do that same null test. I guarantee it'll be markedly higher than -70dB.

In addition, these plugs are meant (in the default configuration) to be run on every channel in a session. Even if it were -70dB, 100 tracks of it would result in -50dB which is nontrivial noise for many genres.

[ScuzzyEye]

Originally Posted by MonkeyBars

We are speaking here of AirWindows's ConsoleXYZ plugins, which are definitively nonlinear – it's kind of the whole point of using them.

True. Actually, after I posted, I tried my +48 dB theory with Pro-Q 3, and it actually distorts when boosted above 0 dBfs. So even my own test started failing with too much gain.

I'm interested in post-fader inserts for Fiedler Audio's Atmos Composer, which is entirely non-linear, but a boost before it would throw off the entire mix, because the track fader is removed from the processing because it has its own internal send. So this work-around wouldn't help at all for what I'm doing anyway.

[mschnell]

Originally Posted by MonkeyBars

Agreed, but that's not the case here. We're speaking about subtracting 6.02 dB then applying an effect then adding back 6.02 dB. That is lossy.

This only might be if the effect is non-linear such as dynamics (and does not feature an input level control), but not for linear effects sich as EQ, (pure) reverb, ...

[mschnell]

Originally Posted by ScuzzyEye

Pro-Q 3 ... actually distorts when boosted above 0 dBfs.

I would call that a bug, unless exactly documented.

[Ozman]

I just stumbled across this thread and simultaneously these videos:

https://www.youtube.com/watch?v=Lvo_vt5nG9I
https://www.youtube.com/watch?v=IArUNueVMo0

Now Reaper can have post fader inserts on ANY track!!! AWESOMENESS!

[smandrap]

Now Reaper can have post fader inserts on ANY track!!! AWESOMENESS!

Yes, nice trick... Until you create a pre-fader send and you have the fake post-fader fx in the return channel too.

Useful indeed, but still needs native implementation

[MixMonkey]

Just bumping this Dolby Atmos/Fielder Plugin needs this!

[goateeproductions]

Originally Posted by MonkeyBars

Agreed, but that's not the case here. We're speaking about subtracting 6.02 dB then applying an effect then adding back 6.02 dB. That is lossy.

It's nice to see that people are using this script and plugins. Considering the loss, this can be tested easily with a null test, which I just did, and it revealed some interesting results. The null test consists of:
- Channel 1, white noise generator, routed to channels 2 and 4.
- Channel 2, post-fader-start -> ConsoleLA channel -> post-fader-end, routed to master
- Channel 3, folder track with ConsoleLA channel, routed to the master
- Channel 4, child track to channel 3 (so, we have real post-fader send), polarity inverted

Now you can test the null by applying the same amount of volume change to channels 2 and 4. Results:
- unity gain -> perfect null
- -60dB -> perfect null
- +12dB -> perfect null
- +5.01dB -> -114dB null

This test revealed that the volume change the plugins do is not 100% accurate compared to the volume change with faders. Probably a rounding error. However, the system doesn't create any loss. Not more than Reaper faders themselves.

Let's compare the true post-fader situation with one created by the plugins with -10dB fader gain.

True post fader:
Signal -> fader -10dB -> plugin

Post-fader-script:
Signal -> gain -10dB (post-fader-start) -> plugin -> gain +10dB (post-fader-end) -> fader -10dB

Since Reaper has a 64-bit engine, it really doesn't matter if you have the first plugin doing -100dB gain and another +100dB. It's still null, no matter what the original audio is.

[MonkeyBars]

Originally Posted by goateeproductions

Since Reaper has a 64-bit engine, it really doesn't matter if you have the first plugin doing -100dB gain and another +100dB. It's still null, no matter what the original audio is.

This is getting to the edge of my knowledge pool, but what you seem to be saying is that technically speaking there is resolution loss here, but the 64-bit engine ensures that the loss stays so far below audible that it's inconsequential... but once again, what if there are 500 tracks, as in a large orchestral recording? -114dB noise on each would add up to -87dB which you could argue is audible.

So it's probably not an issue, but is indeed lossy at certain calculations (i.e. level changes).

[goateeproductions]

Originally Posted by MonkeyBars

This is getting to the edge of my knowledge pool, but what you seem to be saying is that technically speaking there is resolution loss here, but the 64-bit engine ensures that the loss stays so far below audible that it's inconsequential... but once again, what if there are 500 tracks, as in a large orchestral recording? -114dB noise on each would add up to -87dB which you could argue is audible.

So it's probably not an issue, but is indeed lossy at certain calculations (i.e. level changes).

There is no noise or loss. This experiment does not null perfectly with some volume settings because of the rounding error of the volume adjustment. I could probably fix it, but I don't see any reason for it. In the situation where it doesn't null perfectly, I could adjust the volume of one of the channels by less than 0.001dB, and the system would null again perfectly. If someone really has (not theoretically thinking of) a real-world use case where this kind of volume difference matters when using the script, I'll be more than happy to fix this. But there is no noise or nonlinearity. It doesn't matter if you have a million tracks using this system. The worst thing that could happen is that your mix is out of balance by less than 0.001dB compared to the theoretical result. And since it's a linear volume error of a single track, this error does not multiply by the number of tracks. If you boost a million tracks by 0.001dB the sum is still just 0.001dB louder. I hope this answers your concerns.