Timing issues when exporting tracks and busses in Logic Pro.

[Beardy]

Hi folks,

I have a recurring issue when bouncing out stems for mixing and I'd love to know if I'm either doing something wrong to if it's just "the way it is"?

In my Logic sessions I always use many busses for grouping and for effects. I'd say I regularly have up to 30 busses per song project. And somewhere round a hundred tracks of audio and soft synths.

All is fine until I come to export the tracks and busses as stems for an external mix engineer to work with. I've found the stems from the busses tend to be out of time with the straight audio track stems. I was first alerted to this when I compared the dry effect buss stems - for example a parallel compression bus. If I line up the kick drum stem with the parallel compression stem at bar 1 beat 1, they're simply not in time. I've found this to be the case with the other buss stems too. Less noticeable with wet effects like delay etc but immensely frustrating at the same time.

I've experimented using single drum tracks and parallel comp buss stems - same problem. I tried rendering one bus at a time and still all to no avail.

Settings wise I have the Plug-Latency compensation set to "All".
I do use a lot of UAD plug (2xSatellite Octos) and wondered if they were responsible?
I'm using a 32GB 2108 MacBook Pro so pretty powerful.

Anyone any ideas out there? Is this a well known Logic DAW flaw? I've tried finding similar complaints form other people but found any solutions yet. It's driving me crazy!

Help me Desmond wan Kenobi, you're my only hope!

Thanks in advance,

Beardy

[Jack Ruston]

As a mixer, frequently recieving sessions from logic, I'd say that 70-80% of the time there's a weird timing or other arrangement artifact in the bounces. I've mixed tracks and found the arrangement curious, then to find that it had put things in weird places, or that some files were late etc etc. It does seem to have quirks of operation that catch people out, but it's also prone to buggy behaviour.

[muzines]

Does this only happen for you in heavy/complex projects, or can you reproduce it simply?

Ie, start from an empty project and set up a simple drum track with with paralleled compressed stems, no UAD, and bounce out the stems. Do you still get timing issues?

Do the same, but with UAD processing this time. Do you get these timing problems?

[CS70]

For what it's worth, a mac-using friend often complains of timing discrepancies when he imports the tracks I made for him together with his own. I never understood exactly how but
maybe it's something related?

It seems a very odd thing, as using busses that way is such a basic way of any workflow that if there were consistent timing bugs between tracks and buses I would expect they'd been discovered already? Just wondering.

[Hugh Robjohns]

Interesting... is this related to plugin delay compensation? Exported tracks (with plugin effects) not taking it into account when exported, and receiving systems not knowing to realign ?

[muzines]

For what it's worth, a mac-using friend often complains of timing discrepancies when he imports the tracks I made for him together with his own. I never understood exactly how but maybe it's something related?

The problem with these kinds of things is while the problem can often be seen easily enough, it's often hard to pinpoint exactly what's *caused* the problem. These days projects are often hugely-complex, multiple chains of third-party plugins and complicated routing. Plus, people make mistakes, hit keys they didn't mean to, click-and-drag things they didn't mean to, and invariably mess things up without realising, or have preferences set they weren't aware of which affect behaviour.

The thing to try is to simplify, and troubleshoot like I suggested above. If those situations show no obvious problems, then the issue isn't a simple, easily reproducible issue, which means that far fewer people will hit it, or even notice.

For example, let's say hypothetically there is a PDC bug, but it only shows up if you route a track through more than 4 busses in a chain, alongside a parallel bus, using a sample rate of 96Khz, when there is one of a small handful of particular UAD plugins on the track. That combination of factors is going to affect a small amount of people a small amount of time, which is why these kind of bugs can be hard to spot.

Try the simple things first to see whether it's something that's easy to check, and if those things perform ok, then you can perhaps take a copy of the complex project, and strip it down until you have the essence of the problem. At this point, you might be able to recreate it, and find whatever's causing the issue (be it a bug in Logic, a bug in third-party plugins, a combination of the two, a settings issue, or whatever)

It seems a very odd thing, as using busses that way is such a basic way of any workflow that if there were consistent timing bugs between tracks and buses I would expect they'd been discovered already? Just wondering.

Indeed. I'm not seeing a huge amount of complaints about these kinds of timing issues on the Logic forums, at least related to the OP's issue, but that doesn't mean there aren't any. If we can find a cause, then it can be filed as a bug report and hopefully addressed, if it's something that needs to be fixed.

I'm not really able to do exhaustive tests at the moment, but I did a quick run though with a paralleled bus, the bus having a Logic compressor on it, and doing a bunch of bounces, exports, bringing the bounces back in and null testing them, and everything I tried nulled as expected, apart from the compressor bus which didn't *quite* null (perhaps because of the compressor settings and soft-clipping used?). In this case, I don't think the compressor adds latency anyway, so not necessarily a good test, but it was just a quick thing I tried. I'll try another look with a UAD plugin and see...

Like I say, I can't really get into the weeds on this at the moment, and spend time investigating multiple routings, plugins and sub-millisecond timings and so on, but if the OP can see if he can reproduce the issue, I'm happy to try and confirm and verify any findings...

[muzines]

Ok, I did a quick test with a UAD SSL Channel Strip as the parallel compressor, 25.6ms latency (Logic reports the latency a plugin uses if you hover over the plugins slot).

Again, there were no timing differences (verified down to sample accuracy) when I compared a bounce in place version of the source track, compared to exported versions of the source track and the aux channel with the plugin processing on - the delay compensation was working as expected in this (simple) example.

[CS70]

Plus, people make mistakes, hit keys they didn't mean to, click-and-drag things they didn't mean to, and invariably mess things up without realising, or have preferences set they weren't aware of which affect behaviour.

Yes, I always chalked it down to that.

My friend knows little about DAWs, even less about audio but somewhat lacks the confidence to acknowledge it and learn. As the tracks otherwise work fine, I've always thought it was down to user unawareness.

[Beardy]

Guys - thank you for the detailed responses and interest in this. I’ll try what you said Desmond. Sounds like you’ve tested already but I’ll experiment with different routings and various levels of complexity and report back.

Your comment about multiple routings creating an effect resonated (no pun!). I do like to feed tracks and busses into each other a fair bit!

One thought I had was after watching a video on rendering busses as audio - simply set up some blank tracks and set the inputs as the busses and then record the buss outputs in real time. Should work?

Anyway will report back when I get chance. I’m away on your right now so not getting a lot of down time. Thanks again!

Beardy.

[muzines]

One thought I had was after watching a video on rendering busses as audio - simply set up some blank tracks and set the inputs as the busses and then record the buss outputs in real time. Should work?

You're better of doing bounces/exports, imo. Firstly, you can do it offline and thus much faster, and secondly, if you real-time record from busses back through to inputs, you need to check your audio interface latency and recording offset, in conjunction with the overall delays due to PDC aren't changing the timing. I haven't checked the current behaviour of this for a while, but certainly this would often affect timing in older versions.

I'd do a few quick tests on a test project to check the timing of recording if you really want to record busses/stems in realtime (I don't really see the point in doing it this way, unless you're running stuff out to outboard gear and therefore need to record in realtime.)

[Beardy]

OK. Good advice. Thanks Desmond. I'm about to have a little experiment. Let's see.

[imagine1957]

I am having the same issue and it is making collaboration with others almost impossible. I just exported 44 stems and the midi for kick was exactly on the beat, but the exported stem had it off by about 1/16 of a beat. I saw that you referred to a video about busses above .. can you link that video please? Thanks.

[coove]

Hey! I was having a similar problem, I turned off "Software Monitoring" on Preferences/Audio/General and that seemed to do the trick.

[BenWilesMusic]

I've never noticed this problem, and I use stems quite a bit. But that might be because the way I do it.

First I (usually) print (bounce tracks in place) all virtual instrument tracks so I'm just working with audio tracks. (That's just to prevent any buggy or inconsistent behaviour in VIs or bus routing - which used to be a bit buggy in Logic - from causing head scratching later on.)

I then create tracks for all my stem aux busses. I bounce each of these aux tracks in place (again using bounce tracks in place, and mute the original busses.

At this point I would be able to identify any timing problems, because all the stems are lined up in the project. (But I've never had any.)

Final stage is to trim all the regions for export, making sure they all start at the same point.

I then export the stems by exporting the trimmed regions.

Are you doing something similar, or are you working differently?

[BenWilesMusic]

Btw, you can create an arrange track for an aux by selecting the aux in the mixer and pressing control-T.

[ManFromGlass]

That is a good practice but when the deadline clock is ticking I usually don’t bounce any synths or other plugs including FX to audio tracks. I’ve been doing a one pass “bounce” of up to 8 stems at once for many years. I also work in the box and use UAD plugs.
Years ago I remember having timing issues doing the proper bounce not in realtime.
What I mean by “bounce” (in quotations) is that I’ve got Logic set up to record up to 8 stems in one pass using AutoPunch.

I haven’t tested timing in awhile and I would say my projects never come close to the number of tracks Beardy uses. But I’ve never had any comments from show mixers I’ve worked with.

I recall there was a YouTube done by a German fellow on a drumbeat showing how different settings of Logic’s compressors cause latency. This was years ago so I don’t know how relevant it would be to todays version of Logic.

[Beardy]

Hey folks fingers crossed the latest 10.6.3 release might address this situation...

Fingers crossed!

https://support.apple.com/en-us/HT203718