Plugins / Delay Compensation / Timing / Downbeat [PT-263943]

[JingleDjango]

Yeah, he and I had the same conversation. Guy's a saint. This is the type of communication we need from Avid.

[JingleDjango]

I suspect there is no delay compensation whatsoever for MIDI events. Does that seem right?

Avid finally responded to my support case by telling me I should remove any delay inducing plugins on both the MIDI trigger track and the tracks receiving the key input. We already know this is ineffective because there could be any number of delay inducing plugins in the path between those two tracks.

I did another experiment of sync'ing Ableton Live to PT using MTC through the macOS IAC device. The transports of the two DAWs play perfectly in sync until delay inducing plugins are activated at any stage of the signal path (in this case Avid's Maxim).

[XJENSEN]

So they admit this critical functionality is broken in Pro Tools.

[JingleDjango]

I don't know if they'd phrase it that way but that seems to be the implicit message they're sending.

[bolooki]

so basically we need to shift the midi being sent to LFO tool earlier by XXXX samples as we go... whats the fastest way to find this number?

A reliable work-around would be fine. We can't rely on Avid to fix this, if at all bc if they did it would affect how your mix sessions sound in later PT versions I think. (ie the midi-based fx and LFO tool's out of sync that you dealt with originally would sound different if this were corrected in PT, right?) it'd have to be an option or something.

[RTGraham]

Originally Posted by bolooki

We can't rely on Avid to fix this, if at all bc if they did it would affect how your mix sessions sound in later PT versions I think. (ie the midi-based fx and LFO tool's out of sync that you dealt with originally would sound different if this were corrected in PT, right?) it'd have to be an option or something.

I was giving this very issue some thought last week.
From my own perspective - and I can't speak for anyone else on this, but this is what makes sense to me - because the degree to which the timing is "broken" in a session is related, perhaps more complexly than simply, to the combination of delay compensation factors in a session, and because that "brokenness" can change as a result of something as simple as tweaking the mix or changing something about the routing, I'm not particularly concerned about how a proper fix might change the sound of an existing session when it's reopened.
I've already had that bullet jump up and bite me. Under the current circumstances, it's already very very possible to open a session and inadvertently change the timing of the dynamic processing or rhythmic gate processing or even the musical elements themselves, and it can be so subtle as to not necessarily notice it at the time but then wind up with phase issues that get magnified in mastering. The whole thing already feels unpredictable and random in practice, so "breaking" existing sessions isn't - to me at least - that great of an issue. They're already broken.

So I'd rather see this properly fixed, as opposed to trying to figure out what the predictable workaround(s) might be.

[RTGraham]

Originally Posted by JingleDjango

Here is one more short video detailing the ADC bug concerning LFOTool:
https://www.youtube.com/watch?v=c2Xfba9xgg0 (~50sec)

I've confirmed the same behaviour occurs if I downgrade to PT ver. 2020.03.

The gist of it is that sometimes LFOTool will cause a track's delay compensation value to null. Disabling the insert will return the track ADC value to its correct state.

RTGraham, I'm curious whether you've noticed any of the same behaviour on your system or with any different plugins.

I've finally just now watched these two videos. I hadn't previously noticed that, and it's an interesting aspect of this.
I'll have to try to find some time to test whether I'm seeing the same behavior here.
BUT - in the meantime, a couple of immediate thoughts.
I'm under the impression that in general practice, Pro Tools has evolved to try to give us the ability to use virtual instruments in real-time with minimal lag, even when the session otherwise contains processing that requires delay compensation.
One of the ways PT attempts to do this for us is by not compensating a track that's meant to be a real-time MIDI record or trigger track - i.e. an instrument track, hosting a virtual instrument, routed directly to the hardware outs, and either record-enabled or receiving MIDI from a MIDI track that's record-enabled.
So even if other tracks / busses / master paths are causing and using compensation, it's generally possible to still use a VI with minimal latency as long as you get the routing right.

WHAT IF what's happening is that PT is perceiving that an audio plugin - something rhythmic that requires timing information like Filter Gate or Pumper or Boom or LFOTool or Tremolator - is essentially MIDI-enabled, and as such is treating it like a virtual instrument, and attempting to give us real-time playability by bypassing the delay compensation for whatever track it lives on?

I don't know if that's what's actually happening. And if it is, it's an interesting task to try to differentiate when the compensation should be applied and when it should be bypassed.

[RTGraham]

Originally Posted by bolooki

sup y'all - got a response from Steve himself from Xfer concerning LFO tool:

https://xferrecords.com/forums/lfo-t...aax#post_74810

"steve_xfer 2 days ago:

AFAIK This is AVID's bug to fix, there is nothing that LFOTool is reporting, it recieves time position info from the DAW which is being reported to LFOTool incorrectlly.

My understanding is it affects other plugins beyond LFOTool.

The root issue might be more complicated, but my experience was that ProTools is not compensating time position sent to plugins properly when you have a latency greater than 2048 samples."

2048 is a good hint here I think as far as knowing when to give this issue a heads up or at least when to freeze before moving on.

Interesting.
In my simple testing with Filter Gate, it's possible to break it even with delay compensation amounts below 2048.
One instance of Maxim appears to induce 1024 samples of latency, according to the readout on the track.
When I print the Filter Gate affected track in order to measure by just how much the early audio that I'm hearing is early, the resulting audio is ahead of the grid by 1020 samples.
With two instances of Maxim - so 2048 samples of latency - the resulting print is early by 2043 samples.
So allowing for a bit of minor skew depending on where in the source waveform the Filter Gate happens to catch the audio (zero crossing vs. peak), the issue appears to be directly proportional to the amount of compensation.
What I haven't had time to do further testing on yet, because I think it will be complex and convoluted and tricky to come up with a good methodology for, is figuring out how compensation in *other* paths in the session can or can not affect a path with a rhythmic timing-dependent plugin that's outside of the compensation-inducing path.
So far I've just been limiting this to compensation in the same path, prior to the processing in question.

[JingleDjango]

I've been busy this week and haven't have much time to spend on this, but my suspicion is that the main issues here are:

1) lack of side-chain delay compensation
2) lack of automatic delay compensation for MIDI events

1) JFreak will be around to tell us that side-chain ADC is an Ultimate-only feature but. . . I've worked extensively on HDX systems and in my experience side-chain ADC was only reliable if all of the inserts in the signal path were DSP only. AAX DSP is not widely supported and it is impractical to try to mix only using those plugins if you have a preferred toolset that includes native plugins. If SC ADC were properly implemented across the Pro Tools line I wouldn't even bother with LFOTool. I'm comfortable dialing in the pumping effects I want using a compressor or a gate.

2) Avid's response is that I should manually adjust MIDI offset so I've asked them to confirm that there is no form of MIDI ADC yet in Pro Tools.

I understand where bolooki is coming from regarding old sessions. If MIDI ADC (or whatever it takes to fix this) might have the effect of altering playback on older sessions then all it would take to address that is an Options menu toggle like the one we already have for plugin ADC.

[Franklyn]

Originally Posted by JingleDjango

Thanks for sharing your experience, Franklyn. I've been assured by a 3rd party developer that the issue lies beyond their control. With no input from Avid on this I have no way to judge that claim but it sounds as though RTGraham has had a similar conversation with a different dev sharing the same opinion.

I don't have access to any Toontracks plugins but I did just try a test session following the same methodology as RTGraham outlines in the original post, but replacing the SigGen & Filter Gate plugs with an instance of Boom. As instances of Maxim are added in series before Boom the drum pattern goes out of sync in the same way we experience with the other inserts. You should try yourself.

Yes, the problem I described is not the one you described.

1. what I described:
internal plug in sequencer ADC is not correct implemented by many 3rd party instruments like Kontakt, Maschine, Komplete Controll, BFD, Addicted Drums. And I am sure thats 3rd party issue (because other ones like toontrack does it right). Those Instruments internal sequencer are going out of sync also when they are inserted as 1st plug in on an instrument track.
so, sorry, this is not what this threat was for.

2. what you described in this treat:
bpm synced fx plug ins like LFO tool that going out of sync with the downbeat is definitely an issue on Avids side.

This issue should be corrected with very high priority. Also if only a few people realized this. This makes also a lot of mixes worser for people how havent found this out. timing and sync is the most important thing in modern music production and I had the exact same problem in my last singlemix. LFO tool and also waves beat synced effects are complete out of sync.

adjusting them manually is not a working solution because every time you add something to the mix that changes the over all ADC you have to manually adjust every beat synced plug in again. ?!?

Also backward compatibility is not a argument.

I never heard that Sidechain ADC is a ultimate or HDX only feature. I work with PT ultimate but not with HDX. And I have this problem.

In Avids official comparison list of all Protools Versions is no difference in ADC between the PT versions.

I hope so much that Avid will react soon to this.

Thanx a lot RTGraham for the good description of this issue.

Frank