Fader latency - response time dependent on buffer size?
 

Hi,

I've had issue for a while...

I've had an Avid Artist mix (brand new in box) for around 4 or so months now and something I've noticed, which really bothers me is the speed in which ProTools responds to fader moves seems dependent on playback engine's buffer size. For instance if I push a fader up from -∞ to -0db with a buffer size of 1024 samples, there is a significant delay (less than a second, but noticeable) between when that fader is physically moved to when I hear the volume of the audio change. At 64 samples it's far less noticeable, but it's certainly no analog fader.

When I'm writing automation things get even more interesting...
Keeping in mind that when a session gets quite dense (towards the end of the mix typically) it's not uncommon for me to be mixing with a higher buffer size selected to avoid errors. When I write automation I've noticed that i have to anticipate quite a fair amount for me to make the change I want for me to hear it in realtime. Annoyingly though once I hit stop, when I playback the automation I've just written to that track it sounds like it's compensated ahead in time even more!

An example to illustrate what I'm talking about - let's say a chorus is coming up and I'm going to do a big ride on the master fader for it, I'll anticipate that it's coming and turn up the fader intentionally ahead in time to make sure I catch the downbeat. Sweet? No. When I play it back that volume ride will be boosted even earlier than what I heard when i was writing it in the first place.

If I write automation with a mouse in ProTools on the fader it seems to be fine and I don't experience this problem so it has to have something to do with the avid artist mix.

Am I crazy?

I'm on Macbook Pro Retina 2014 (Yosemite), ProTools 11 standard.

Thanks

Re: Fader latency - response time dependent on buffer size?
 

Interesting that nobody has commented on this. I also notice this on both of my native systems with and S3 and running HD software. The way the moves get compensated means to make a proper mix, I have to make moves that sound "wrong" but feel right, then they play back at the right spot (where I moved the fader).

I've been able to deal with this in my own mixing, but when I have clients with me, they often think I'm missing fader rides, then I have to explain the compensation situation to them.

I'm going to do some testing tomorrow on an HDX system to see how that behaves.

Re: Fader latency - response time dependent on buffer size?
 

Don't have this issue (on HDX). When I move the fader I can see the automation writing exactly where I want at the same time I'm moving the fader. Same as it would be if I was to do it by mouse.

As for buffer/latency -- that of course will affect when things come out the speakers depending on how elaborate your b-chain is but Pro Tools is "visually" compensating for what it can with the exception of using the scrubber (but I'm not doing automation if I'm using the scrubber)

So I guess my question would be: if you are watching the automation as your write, are they happening at the same time?

Re: Fader latency - response time dependent on buffer size?
 

and if not, could it be lagging graphics? In any case, I have never felt or noticed this on my rig either(HD/Native) whether pushing faders on my Artist Mix or Faderport(the original single fader from Presonus):o

Re: Fader latency - response time dependent on buffer size?
 

I just tested this on a 12-core trashcan MacPro running HDX. It seems the buffer size definitely affects automation latency. I tested with Mute automation and putting a mic up to the mute button as I pressed it. This gave a nice impulse to measure the automation trigger time with. I performed the test using a copy of a completed 90 minute TV mix, and once again with an empty session inserting 1 new audio track. Both sessions are 24bit 48kHz.

The test that shows automation is delayed and the delay depends on the buffer setting and session complexity.

In test #1, with my mixing template, the automation is about 84 ms late with a buffer of 256 samples and 168 ms late with a buffer of 1024.

Test #2 shows 26ms late with a buffer of 256 and 52 ms late at 1024.

I performed the test several times and readings were fairly within 10ms each time. I'm uploading 2 screenshots of the readings here.