Thoughts on Negative Delay Compensation?

Discussion in 'Workflow Tips & DIYs' started by marclawsonmusic, Feb 26, 2019.

  1. marclawsonmusic

    marclawsonmusic Senior Member

    1,282
    294
    Mar 16, 2012
    Florida
    Hi all,

    I saw this video from John Powell and was honestly shocked at how super-neat and quantized his MIDI is:



    In the comments, he mentions using 'Negative Delay Compensation on all tracks' to achieve this. Makes sense, but doesn't that mean the audio will play right on the beat? In his video, there is clearly an audio delay as the playhead moves through the notes...

    Beyond that, any thoughts on using Negative Delay Compensation? It would be helpful to be able to quantize MIDI if it still sounded good... I also think this would make it much easier to go from DAW-to-score... and also easier to orchestrate in the DAW - copying and pasting parts becomes less of a problem with this.

    Thanks in advance,
    Marc
     
    Divico likes this.
  2. Alex Fraser

    Alex Fraser Senior Member

    1,090
    1,509
    Jun 21, 2017
    UK
    For some reason I can't play the video..
    Like you say, it's probably a great and quick way to work if the final result is to be played by an orchestra. If you're writing a track with the final product being VI output, I'd bet that an "off grid" workflow would ultimately result in more realism. But I could be proven wrong - it's not for me to question JP!
     
    Last edited: Feb 26, 2019
  3. OP
    OP
    marclawsonmusic

    marclawsonmusic Senior Member

    1,282
    294
    Mar 16, 2012
    Florida
    John Powell's MIDI mockups sound better than any I have ever heard, and he is using most of the same tools that everyone on here is using (He publishes his template annually on social media and posts these MIDI mockups all the time).

    So, it's not just about going DAW-to-score - it's about getting a great result inside the DAW too.

    Here are some URLs in case they work better... EDIT: Looks like the forum converts these to media links too... oh well.


     
    Alexandre likes this.
  4. brek

    brek Active Member

    154
    143
    Oct 30, 2016
    One other great reason to quantize - conforming a large cue to a new picture edit when every note starts slightly ahead of the measure. :crying:

    This is one of those old "rules" that has some truth to it, but is a bit overplayed - particularly with modern VIs. Personally, I think note dynamics and durations/transitions are often more likely to give away the realism.

    @NoamL has some good demonstrations of realistic quantization using micro tempo adjustments that are very effective.
     
    goalie composer likes this.
  5. David Chappell

    David Chappell Active Member

    150
    140
    May 19, 2013
    I always write entirely on grid using negative delay, 100% quantized, solo instruments being the only exception.

    At first glance it seems like this is a recipe for robotic, lifeless playback, but you have to remember that midi notes are really just playing back audio files, so what the midi notes look like is fairly irrelevant. They key here is that humanisation already exists in the samples. An ensemble of players playing a short articulation, for example, won't be playing exactly on the beat. Some will be slightly before, some slightly after, by only a few milliseconds, and this will vary for each note, RR, velocity, recorded. So when you press a MIDI note what you're really playing back is an audio file where all of the players are slightly off beat each time. This is the humanisation. Having each midi note be slightly off beat means that you're taking a sample that already has variations in timing baked in, and giving it further variations in timing.

    Since some players will play slightly before the beat, developers need to cut the samples a few milliseconds before the beat as well. CSS for example cuts the shorts 60ms before the beat. Then it's just a case of putting the note exactly on grid, and telling the DAW to actually play that note 60ms earlier, so that the midi note plays back the sample file with the "peak" landing right on the beat.

    Much more important for realism is good velocity/ CC management, and automation of the tempo track (still keeping everything 100% quantized).

    At least, this is my reasoning that I worked out for doing it this way :)
     
  6. OP
    OP
    marclawsonmusic

    marclawsonmusic Senior Member

    1,282
    294
    Mar 16, 2012
    Florida
    FWIW, NoamL's Thanos script is what got me thinking about this in the first place!

    And, I typically use a played-in piano guide to create a tempo track, so that helps with the music 'breathing'. Hmmm... good food for thought here.
     
  7. goalie composer

    goalie composer Active Member

    173
    80
    Oct 21, 2013
    Would love to check these out. Do you have a link to a video of this handy?
     
  8. OP
    OP
    marclawsonmusic

    marclawsonmusic Senior Member

    1,282
    294
    Mar 16, 2012
    Florida
    I quit using quantisation back in the earlier days of EWQL. Maybe samples have changed enough to where this isn't so much of a problem any more. Certainly makes it easier to edit things.
     
  9. brek

    brek Active Member

    154
    143
    Oct 30, 2016
    goalie composer likes this.
  10. babylonwaves

    babylonwaves Senior Member

    878
    546
    May 9, 2015
    me too. can't think of doing it differently. i might loosen up the quantisation in certain cases but i always adjust every single instrument in my template with negative delay to make sure i can move around regions without having to adjust the timing.
     
  11. Alex Fraser

    Alex Fraser Senior Member

    1,090
    1,509
    Jun 21, 2017
    UK
    That's really interesting and I can't quite believe after so many years I hadn't really considered it that way. Old habits die hard I guess. I feel some experimentation coming. I guess it also depends on how consistent the library is with regard to sample editing. Maybe that's why everyone loves CSS.. ;)
     
  12. brenneisen

    brenneisen Active Member

    227
    214
    Feb 23, 2017
    those are not mockups, live performances instead.
     
    karelpsota likes this.
  13. shomynik

    shomynik Active Member

    332
    120
    Aug 29, 2016
    Germany
    Was thinking the same. Don't know as a fact, but strongly believe those aren't samples.
     
    karelpsota likes this.
  14. Vonk

    Vonk Member

    64
    34
    Nov 17, 2018
    How are you calculating the negative delay? it vary between libraries? - between articulations? I find I'm having to use them, but am unsure of the best approach.
     
  15. OP
    OP
    marclawsonmusic

    marclawsonmusic Senior Member

    1,282
    294
    Mar 16, 2012
    Florida
    Are You sure about that?
     
  16. brenneisen

    brenneisen Active Member

    227
    214
    Feb 23, 2017
    it is.

    Screenshot_20190226-155229.png
     
    shomynik and marclawsonmusic like this.
  17. OP
    OP
    marclawsonmusic

    marclawsonmusic Senior Member

    1,282
    294
    Mar 16, 2012
    Florida
    Well damn! No wonder it sounded so good. Can I delete this thread now? Hahah

    PS - Thanks for clearing this up.
     
    brenneisen likes this.
  18. babylonwaves

    babylonwaves Senior Member

    878
    546
    May 9, 2015
    i play a short sample. bounce it and measure the delay in the sample editor. or, if i'm lazy i just push the delay until it suits my needs. of course this is all under the assumption that the articulations within the instrument have the same delay. but it works pretty good in most cases.
     
    Vonk and goalie composer like this.
  19. NoamL

    NoamL Winter <3

    2,437
    4,986
    Jul 6, 2015
    Los Angeles
    where is this comment about using negative delay? I quit facebook a while ago.

    v busy day at work but I have some thoughts on this wholetopic ;)
     
    Karma, chillbot and marclawsonmusic like this.
  20. OP
    OP
    marclawsonmusic

    marclawsonmusic Senior Member

    1,282
    294
    Mar 16, 2012
    Florida
    upload_2019-2-26_16-14-51.png
     

Share This Page