|
|
|
04-17-2021, 06:44 AM
|
#1
|
Human being with feelings
Join Date: Aug 2006
Location: Berlin
Posts: 11,817
|
Volume Automation Recording Render-ahead Latency compensation
The problem is illustrated in a simple little project.
https://www.dropbox.com/s/dxscjrobhl...v6_27.zip?dl=0 (8 MB)
Update. Here's a screenshot of that session.
Preferences/Buffering
Anticipative FX processing set to ON
Render-ahead now defines the latency with which it is actually applied to the track,
during recording of track volume automation,
This does not happen on playback.
A higher render-ahead gives you both a worse response time of the control surface input to actually hearing those changes, but also gets you better performance.
Because of this I've lowered it to 100ms to at least have some form of compromise. My ASIO buffer is at 256 samples, which works well on almost all projects (6700K CPU).
But the crux remains.
What you hear during automation recording, is NOT WHAT YOU GET.
RequestShift the volume automation(or any automation?) during live recording of it with the render-ahead value, so we continue to get what we heard during automation recording.
Or otherwise fix the problem. This latency of input versus hearing changes is already bad, but compensating for that in your head is just about as great if not worse as recording precise drum parts with a 2048 buffer, or in my case 0.1*48000 samples.
The main difference to audio recording is, you can't even rely on what you hear during recording right now.
Last edited by airon; 04-17-2021 at 09:18 AM.
|
|
|
04-18-2021, 02:51 AM
|
#2
|
Human being with feelings
Join Date: Aug 2006
Location: Berlin
Posts: 11,817
|
Just out of curiosity, is the explanation good enough ?
I don't want to fall in to the trap of assuming everyone thinks the way I think.
TLDR Your recorded fader movements affect the material with a latency that is set in the Preferences/Buffer/"Render-ahead" box. I assume that the reason is that the volume envelope is processed along with the effects at the same time.
When you record volume automation, you will hear that automation affecting your audio with latency.
When you play back your volume automation, you will hear that automation affecting your audio without latency.
So there's a difference to recoding and playing back volume automation. That cannot stand.
So my request is to either compensate for that latency, or fix the problem in another way. For example, volume automation could be recorded with timestamps that compensate for the render-ahead latency, so the playback on the next loop pass sounds right.
How's that ?
Last edited by airon; 04-18-2021 at 02:58 AM.
|
|
|
04-18-2021, 03:51 AM
|
#3
|
Human being with feelings
Join Date: Jul 2009
Posts: 1,231
|
Great explanation. I desperately want this looked into as well, really significant when working with control surface automation passes.
|
|
|
04-23-2021, 01:44 PM
|
#4
|
Human being with feelings
Join Date: Aug 2006
Location: Berlin
Posts: 11,817
|
A bump to kick it in to view for us users and the devs.
|
|
|
04-24-2021, 12:54 AM
|
#5
|
Human being with feelings
Join Date: Jan 2020
Posts: 1,927
|
Yes, this is sorely needed. Your analogy to recording percussive parts with a high buffer is apt—that’s precisely what it can feel like doing rides in REAPER. I really hope for some improvement in this area.
|
|
|
07-13-2021, 03:20 AM
|
#6
|
Human being with feelings
Join Date: Aug 2006
Location: Berlin
Posts: 11,817
|
Not just a bump but also a question for the devs, should you have the time for it.
Moving the volume envelope processing, which is what I assume recording volume automation falls under, would require some major rework ?
And that's the reason this is way further down on the to-do list, if it's even on there right now, right ?
The easier thing right now would to have the volume automation time-stamped, so what I hear when recording it matches what I then listen back to either on the next loop pass or after restarting playback. At least then I can predict things and make accurate fast moves on a fader.
This could be done by doing that time stamping on current_time+ASIO_BLOCK+antipative_processing_time (an amateur guess). Or is just the anticipative processing time that needs to be compensated for ? IN any case the newly recorded volume automation has to be moved to the right.
Edge cases like setting faders before commencing playback in Latch/Write modes have to handled.
Time to clean this up IMHO.
Last edited by airon; 07-13-2021 at 07:56 AM.
|
|
|
12-13-2022, 11:07 AM
|
#7
|
Human being with feelings
Join Date: Oct 2006
Posts: 47
|
Just trying to understand this request ...
Do you get different result in other DAWs?
I'm just asking because the delay in automation moves in problematic in all DAWs, because the audio has to go through the buffer before it gets played back.
I've had no problems with automation as long as you know that you should NOT listen to the automation moves as you are doing them (because they are delayed). This is one big drawback of digital vs analog. In analog, you will hear the changes instantaneously and you can react to them.
In a DAW, you already have to know what you want to achieve before touching the fader. And don't listen to the moves as you are doing them.
Please correct me if I'm misunderstanding. This problem is exactly the same in all DAWs. Or is there a bug in Reaper?
For example, here's a bunch of people complaining about this on Pro Tools: https://duc.avid.com/showthread.php?t=378957
|
|
|
02-13-2023, 08:21 AM
|
#8
|
Human being with feelings
Join Date: Dec 2015
Posts: 324
|
This just drove nuts until i understood what was going on. I was riding to fader and trying to just catch some attack to turn them down. While recording the automation it sounded like i messed up, but then realised that on playback it was ok. Was thinking of posting a bug report and then found this trhead which is exactly the problem i am experiencing.
Turning anticipative FX processing off seems to resolve the problem, which will work fine a lot of the time, but will be an issue with larger/heavier sessions.
i don't think that fixing this by automatically moving youe automation moves is a good solution.
What you do is what you should hear and what you should get. A solution could maybe be to take track volume out of the anticipative fx chain for tracks on which automations are armed?
Would be nice to see this fixed. but at least turning off anticipative fx works as desired.
|
|
|
02-13-2023, 08:30 AM
|
#9
|
Human being with feelings
Join Date: Dec 2015
Posts: 324
|
Quote:
Originally Posted by copperx
Just trying to understand this request ...
Do you get different result in other DAWs?
I'm just asking because the delay in automation moves in problematic in all DAWs, because the audio has to go through the buffer before it gets played back.
|
the issue here is with anticipative fx turned on, then there is 200ms (or value if youve changed the default) delay, an that makes it really unworkable. With that option off that large delay is gone.
The delay due to audio buffer is something else, as you say it's in all daws, it can be small and ok to work with.
|
|
|
04-03-2023, 03:46 AM
|
#10
|
Human being with feelings
Join Date: Aug 2006
Location: Berlin
Posts: 11,817
|
Somehow this latency when recording volume automation seems to be gone now. I tried it with a render-ahead of 1000 ms and felt no latency beyond the ASIO buffer(256) I was using.
But now a few weeks later this no longer seems to be the case. Weird.
I did change from v1.1 to v3 of CSI, but surely that can't be the culprit.
I was under the impression that I need to arm a track to have that anticipative render-ahead not happen. That render-ahead always included the volume envelope, which as an unfortunate byproduct introduces that gastly latency.
I had not armed those tracks during my tests. And now I cannot reproduce that. Oh well. Back to basics.
Last edited by airon; 04-03-2023 at 03:52 AM.
|
|
|
04-03-2023, 08:03 AM
|
#11
|
Human being with feelings
Join Date: Aug 2006
Location: Berlin
Posts: 11,817
|
I suspect this may be in the dev versions, not the main releases right now. Still need to test that though.
|
|
|
04-03-2023, 09:23 AM
|
#12
|
Human being with feelings
Join Date: Jul 2006
Posts: 12,480
|
|
|
|
Thread Tools |
|
Display Modes |
Linear Mode
|
Posting Rules
|
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts
HTML code is Off
|
|
|
All times are GMT -7. The time now is 08:26 AM.
|