|
|
|
03-18-2016, 11:12 PM
|
#1
|
Human being with feelings
Join Date: Dec 2014
Posts: 543
|
MIDI note position
Why does the indicated position of this note change by one tick after I type in a precise value? Other notes in the sequence are doing the same thing. I set the, to 240 ticks, and after applying the edit, the revert to 239. Sometimes they revert as soon as I hit apply, other times it's after I re-select them. It's not a lot, but why don't they stay where I put them at?
Another example.
Even with snap on, when I drag the note start to the grid, and can feel and see it snap, the indicated values are off by one tick. And when the item starts right on the grid line, the first note in the item is still indicated as x.x.959. How is that even possible?
Last edited by panicaftermath; 03-19-2016 at 12:12 AM.
|
|
|
03-19-2016, 09:15 AM
|
#2
|
Human being with feelings
Join Date: Nov 2015
Posts: 1,566
|
Note position bug confirmed as described.
There have been similar bugs with positions before, if your work/projects require this level of MIDI reliability and precision, you'd probably want to look into other DAWs. I opted to change my workflow to avoid editing MIDI clips and notes (fine for my particular style of music), perhaps MIDI issues will be addressed in the far future but it appears the development efforts are currently directed to other more "pressing" things.
|
|
|
03-19-2016, 09:56 AM
|
#3
|
Human being with feelings
Join Date: Jul 2006
Posts: 12,482
|
Confirmed here too (5.20pre25). But doesn't seem to happen with all values. Some work, some don't.
FIXED (5.17)
Last edited by Dstruct; 03-24-2016 at 02:17 PM.
|
|
|
03-19-2016, 01:29 PM
|
#4
|
Human being with feelings
Join Date: May 2012
Location: Brighton
Posts: 1,989
|
Yeah man, you'll be better off going to another daw. I'm using cubase le now for midi, and the difference is significant, so I'm thinking of upgrading.
|
|
|
03-19-2016, 02:47 PM
|
#5
|
Human being with feelings
Join Date: Jul 2009
Location: Afford Slaughterhouse, FL
Posts: 624
|
Dude... It's 2016, and I guess by now we figured out how to count using computers. This shouldn't be happening.
__________________
It takes an ordered complex system to recognize another ordered complex system.
|
|
|
03-20-2016, 01:33 AM
|
#6
|
Moderator
Join Date: Aug 2007
Location: Italy
Posts: 4,327
|
I think it could be related to sample rate rounding, there was a setting around to enable/disable it, but I'm no longer finding it now...
- Mario
|
|
|
03-20-2016, 03:29 AM
|
#7
|
Human being with feelings
Join Date: Jul 2006
Posts: 12,482
|
Quote:
Originally Posted by mabian
I think it could be related to sample rate rounding, there was a setting around to enable/disable it, but I'm no longer finding it now...
|
No, seems to be unrelated.
|
|
|
03-20-2016, 05:30 AM
|
#8
|
Human being with feelings
Join Date: Aug 2007
Location: Near Cambridge UK and Near Questembert, France
Posts: 22,754
|
Just so I can get an idea as to whether this is OCD at work or a genuine serious problem for every day music composition, can any of you give me a feel for exactly how large a temporal difference we are talking about here, ideally in a meaningful format like milliseconds?
I know you are into real real precision, Mik, but I am talking here about real world differences.
IF it genuinely IS that significant, sure you have a point that is relevant to all users.
Personally I have now accepted that the vast majority of DAW software (which didnt start out as MIDI sequencer) is going to fall short of what I was used to back in The Day & for the most part have been prepared to live with it.
But there is that big ol' "but".....
__________________
Ici on parles Franglais
|
|
|
03-20-2016, 05:42 AM
|
#9
|
Human being with feelings
Join Date: Apr 2014
Posts: 943
|
This Bug arise
( since the last loop-resizing or so fix for Midi Item.. I do not know anymore which REAPER Version it exactly was)
in use of
-Piano Roll Time Base / Project Beats...Time .... together with
item left edge resizing or position without any snap etc. pp
See gif .. once with Source Beats editing and the same with Project Beats
Item at exact timeline position 1.0.000
1.
Item Source Pianoroll timebase / Source Beats -Note On raw position within Item exact
+0 ticks
Note Beat position 1.0.000
all correct ..
2. The same Item Source now with Piano Roll Time Base / Project Beats editing
Item snap to timeline position 1.0.000
The note position shifts from/to
Raw +1 ticks/Note Beat position 1.0.000
Raw +0 ticks/Note Beat position 0.4.959
I use the quantize only for easy see the difference
Note On position 0.4.959 is outside of the timeline which begin at 1.0.000 time 0 ....
This should be really fixed!!
BTW.and the tread should of course within/move to Bug reports.....
__________________
I hope you can understand me? Without german beer my written english is always very bad, with beer it becomes unbearable!.
Less is more! To much limited the own creativity.
Last edited by ELP; 03-20-2016 at 06:10 AM.
|
|
|
03-20-2016, 07:08 AM
|
#10
|
Human being with feelings
Join Date: May 2012
Location: Brighton
Posts: 1,989
|
Quote:
Originally Posted by ivansc
Just so I can get an idea as to whether this is OCD at work or a genuine serious problem for every day music composition, can any of you give me a feel for exactly how large a temporal difference we are talking about here, ideally in a meaningful format like milliseconds?
I know you are into real real precision, Mik, but I am talking here about real world differences.
IF it genuinely IS that significant, sure you have a point that is relevant to all users.
Personally I have now accepted that the vast majority of DAW software (which didnt start out as MIDI sequencer) is going to fall short of what I was used to back in The Day & for the most part have been prepared to live with it.
But there is that big ol' "but".....
|
It makes a difference mate. If samplers trigger events at equal intervals, and notes are not, the sound outcome will be bit different.
@ELP Ah, that's why first hits of many of my instruments sounds different, because the note is shifted 1 tick in front of the item! Solved!
|
|
|
03-20-2016, 09:19 AM
|
#12
|
Human being with feelings
Join Date: May 2012
Location: Brighton
Posts: 1,989
|
Quote:
Originally Posted by snooks
|
Thanks, but I'm using 7 right now, not 8. When I'll be upgrading thought, I'll upgrade to 8.5 anyway.
|
|
|
03-20-2016, 11:06 AM
|
#13
|
Banned
Join Date: Sep 2015
Posts: 1,650
|
Good news, unfortunately they broke automation after 8.0.2. Still, best of luck on your new adventure!
|
|
|
03-20-2016, 12:16 PM
|
#14
|
Human being with feelings
Join Date: May 2012
Location: Brighton
Posts: 1,989
|
Quote:
Originally Posted by snooks
Good news, unfortunately they broke automation after 8.0.2. Still, best of luck on your new adventure!
|
Thanks for heads up. I looked in to it, and it is volume automation issue when it goes below 0db. It's been confirmed and they are working on fixing it. But I must be a really lucky person, that I've got C7 Le and not C8. For me way around that would be to automate instruments volume and not the track itself. Thought, as I wrote at some point, I still might be using reaper for mixing wavs, just not when I need precise midi.
|
|
|
03-20-2016, 02:03 PM
|
#15
|
Human being with feelings
Join Date: Aug 2014
Posts: 11,052
|
Quote:
Originally Posted by ivansc
Just so I can get an idea as to whether this is OCD at work or a genuine serious problem for every day music composition, can any of you give me a feel for exactly how large a temporal difference we are talking about here, ideally in a meaningful format like milliseconds?
I know you are into real real precision, Mik, but I am talking here about real world differences.
IF it genuinely IS that significant, sure you have a point that is relevant to all users.
Personally I have now accepted that the vast majority of DAW software (which didnt start out as MIDI sequencer) is going to fall short of what I was used to back in The Day & for the most part have been prepared to live with it.
But there is that big ol' "but".....
|
Ticks are tempo-based, not time-based. There are 120 ticks in a 32nd note.
This doesn't have anything to do with the preference to land beats on whole samples, does it?
|
|
|
03-20-2016, 04:00 PM
|
#16
|
Human being with feelings
Join Date: Jul 2009
Location: Afford Slaughterhouse, FL
Posts: 624
|
Quote:
Originally Posted by ivansc
Just so I can get an idea as to whether this is OCD at work or a genuine serious problem for every day music composition, can any of you give me a feel for exactly how large a temporal difference we are talking about here, ideally in a meaningful format like milliseconds?
|
Quite the contrary. If the milliseconds difference is not just a display issue and is triggering notes that should be in sync, it is a serious problem that may affect acoustics depending on what the notes are driving and your physical setting.
You would think that a 1ms delay is not a real issue, however it can translate to phasing problems within the 1m range for audible frequencies that are indeed perceptible and in particular could cause problems when mixing with monitors.
This is only the tip of the iceberg. Try moving and shuffling things around several times, in particular using regions, cutting and pasting items or nudging note selections, and the shift of MIDI notes accumulate and start being very noticeable, even visually with notes that lose grid alignment. Of course, you always have the option to quantize, however that usually kills the subtleties of human performance and can again affect the sound of a mix.
The inaccuracies and annoyances of REAPER's MIDI model is one of the reasons it is a weak DAW for note comping.
I wonder how all of this is implemented internally. I see that the API uses 64-bit doubles for positions and adjustments, which should give enough precision for any calculation when executing actions that modify the items. If the results of the calculations are quantized and denormalized properly so that there are no systematic errors affecting the note model, then I don't know where these issues come from.
__________________
It takes an ordered complex system to recognize another ordered complex system.
|
|
|
03-20-2016, 04:15 PM
|
#17
|
Administrator
Join Date: Jan 2005
Location: NYC
Posts: 15,746
|
Quote:
Originally Posted by Dstruct
Confirmed here too (5.20pre25). But doesn't seem to happen with all values. Some work, some don't.
|
Can you post a project where it does it? (working fine here)
|
|
|
03-20-2016, 05:06 PM
|
#18
|
Banned
Join Date: Sep 2015
Posts: 1,650
|
Quote:
Originally Posted by pbk
This is only the tip of the iceberg. Try moving and shuffling things around several times, in particular using regions, cutting and pasting items or nudging note selections, and the shift of MIDI notes accumulate and start being very noticeable, even visually with notes that lose grid alignment. Of course, you always have the option to quantize, however that usually kills the subtleties of human performance and can again affect the sound of a mix.
|
If you look the way MIDI events are positioned (by checking the API and getting your hands dirty with MIDI stuff) you can see that it's sort of impossible for them to gradually lose position by moving items around because their positions are PPQ values relative to the item start. You can move an item 1,000,000,000 times and those values won't change.
Where you place the item is a different matter.
I think some people get tripped up by items not being exactly on grid and leaving snap to media items on or using duplicate actions. Since regions snap to the end of adjacent regions I can see how this happens there too.
|
|
|
03-21-2016, 01:53 AM
|
#19
|
Human being with feelings
Join Date: Dec 2014
Posts: 543
|
Quote:
Originally Posted by Justin
Can you post a project where it does it? (working fine here)
|
Justin, I've attached the project I just used to create this picture. Thanks for your attention on this. Please keep us posted!
It is intermittent, or inconsistent, and does not happen with all notes. I just opened a large multitrack MIDI file in Reaper where everything was locked to the grid. When I copied one of the items from that project into my working project, with the item aligned to the grid, and immediately opened the ME event viewer, it showed all the events off to the left by a single tick. (Which raises the issue that, even if I did the midi in another DAW as suggested above, getting it back in Reaper is still a problem.)
|
|
|
03-21-2016, 02:45 AM
|
#20
|
Human being with feelings
Join Date: Dec 2014
Posts: 543
|
Another screen shot showing bypassing quantizing and unquantizing. It seems from this that Reaper already thinks the notes are quantized.
|
|
|
03-21-2016, 03:57 AM
|
#21
|
Human being with feelings
Join Date: Jul 2006
Posts: 12,482
|
Quote:
Originally Posted by Justin
Can you post a project where it does it? (working fine here)
|
For me it only seems to be happen in "Measures.Beats.100ths" mode.
Project attached.
Try changing the note's position to 1.2.11 or 1.2.12. It will change back to 1.2.10 or 1.2.11 in this case.
5.17pre1
FIXED (5.17)
Last edited by Dstruct; 03-24-2016 at 02:18 PM.
|
|
|
03-21-2016, 04:04 AM
|
#22
|
Banned
Join Date: Sep 2015
Posts: 1,650
|
The "off by a fraction" is a rounding error in the displayed value, not note position.
eg.
- I set the position of a note to 9.1.50
- In Measure.Beats.100ths the displayed value in List View and Note Properties is 9.1.49. In Measure.Beats.MIDI_Ticks it is 9.1.479
- In either case the note position is at 9.1.50 in the Piano Roll.
- Looking at the position with ReaScript it is also 9.1.5, rounding to the nearest 1/10000000000 of a beat.
|
|
|
03-21-2016, 04:39 AM
|
#23
|
Human being with feelings
Join Date: Aug 2007
Location: Near Cambridge UK and Near Questembert, France
Posts: 22,754
|
Quote:
Originally Posted by MikComposer
It makes a difference mate. If samplers trigger events at equal intervals, and notes are not, the sound outcome will be bit different.
@ELP Ah, that's why first hits of many of my instruments sounds different, because the note is shifted 1 tick in front of the item! Solved!
|
Gotcha - never an issue for me because I use either real instruments or VSTis of same.
It just occurred to me that a difference in "note ons" of a few ticks in normal circs would not be audible. See your point.
__________________
Ici on parles Franglais
|
|
|
03-21-2016, 04:43 AM
|
#24
|
Human being with feelings
Join Date: Aug 2007
Location: Near Cambridge UK and Near Questembert, France
Posts: 22,754
|
Quote:
Originally Posted by pbk
This is only the tip of the iceberg. Try moving and shuffling things around several times, in particular using regions, cutting and pasting items or nudging note selections, and the shift of MIDI notes accumulate and start being very noticeable, even visually with notes that lose grid alignment. Of course, you always have the option to quantize, however that usually kills the subtleties of human performance and can again affect the sound of a mix.
|
I would have thought the mere fact that you have sliced and diced several times has already done that.
Certainly that was my experience back when I was only able to use MIDI
__________________
Ici on parles Franglais
|
|
|
03-21-2016, 04:46 AM
|
#25
|
Human being with feelings
Join Date: Aug 2014
Posts: 11,052
|
Quote:
Originally Posted by ivansc
Gotcha - never an issue for me because I use either real instruments or VSTis of same.
It just occurred to me that a difference in "note ons" of a few ticks in normal circs would not be audible. See your point.
|
I still don't really get his point, and it would seem that snooks has confirmed it is just a display discrepancy, and not affecting the audio.
We're talking differences of around 1 millisecond at 60bpm, or 0.5ms at 120bpm, even it it were actually shifting the MIDI. I'm struggling to think what parameters could be audibly affected by that small a discrepancy. Anyway, that's probably for another thread, and it appears that it is a moot point if it is only a display rounding error.
|
|
|
03-21-2016, 04:50 AM
|
#26
|
Human being with feelings
Join Date: Aug 2014
Posts: 11,052
|
Quote:
Originally Posted by snooks
If you look the way MIDI events are positioned (by checking the API and getting your hands dirty with MIDI stuff) you can see that it's sort of impossible for them to gradually lose position by moving items around because their positions are PPQ values relative to the item start. You can move an item 1,000,000,000 times and those values won't change.
Where you place the item is a different matter.
I think some people get tripped up by items not being exactly on grid and leaving snap to media items on or using duplicate actions. Since regions snap to the end of adjacent regions I can see how this happens there too.
|
Yeah, I've never understood the claims of accumulated inaccuracy. It has never happened to me, no matter how much I loop or copy/paste. I think it must be as you say, and it is user error (though, as a few users do fall into it, maybe snap settings need to be a bit more... user friendly?).
|
|
|
03-21-2016, 06:16 AM
|
#27
|
Administrator
Join Date: Jan 2005
Location: NYC
Posts: 15,746
|
Quote:
Originally Posted by snooks
The "off by a fraction" is a rounding error in the displayed value, not note position.
eg.
- I set the position of a note to 9.1.50
- In Measure.Beats.100ths the displayed value in List View and Note Properties is 9.1.49. In Measure.Beats.MIDI_Ticks it is 9.1.479
- In either case the note position is at 9.1.50 in the Piano Roll.
- Looking at the position with ReaScript it is also 9.1.5, rounding to the nearest 1/10000000000 of a beat.
|
I've confirmed this, and also it is worth noting that if it shows you the wrong value and you don't edit it, it doesn't update the underlying values. I'll see if I can fix the display error.
Oh also, in source beats mode the display error does not occur here, or at least is much more difficult to make occur.
|
|
|
03-21-2016, 06:17 AM
|
#28
|
Administrator
Join Date: Jan 2005
Location: NYC
Posts: 15,746
|
Quote:
Originally Posted by panicaftermath
Another screen shot showing bypassing quantizing and unquantizing. It seems from this that Reaper already thinks the notes are quantized.
|
They might already be quantized? Unless you freeze quantization, you can have projects that keep the quantize and unquantized state. Did you quantize it earlier, or record with input quantize?
|
|
|
03-21-2016, 07:28 AM
|
#29
|
Human being with feelings
Join Date: May 2012
Location: Brighton
Posts: 1,989
|
@Judders Difficult to say at this point. I don't know which information goes to synth. But when I was doing that recent example, the difference was even bigger in sound, when there was more midi items. After gluing it got better. But imagine scenario where I have lfo + effects in synth controlled by its step sequencer or pattern. Pattern runs constantly ( doesn't restart on note on), so if the note triggers bit earlier or later you will hear the effect of lfo with ought a sound ( until note on ) or sound of note before the effect. It's marginal, but I don't like that.
A while back I've made apost about first note triggering issue, where the first note always is being played differently then other notes. Another thread was about a glitch where if two midi tracks send signal to same midi ch, when the second track starts, it triggers loads of sounds instead of selected note. There was audio with that example.
But the only issue I had with moving items, and had heard phasing issues on duplicate section, but that was them not snapping properly I suspected. Otherwise I didn't have had shifted notes... I was just abandoning projects.
Last edited by MikComposer; 03-21-2016 at 07:49 PM.
|
|
|
03-21-2016, 08:57 AM
|
#30
|
Human being with feelings
Join Date: Aug 2007
Location: Near Cambridge UK and Near Questembert, France
Posts: 22,754
|
Yeah - what I took from your earlier post was that it WAS cumulative and made worse if you were triggering long samples or sequences of samples.
Judders is thinking the way I was, tiny un-noticeable in audio.... Mik being his usual obsessive self (grin) but its beginning too look like there is at least some effect under the right/wrong circs on the actual audio.
Just popping out for more popcorn and a comfy cushion as this unrolls....
__________________
Ici on parles Franglais
|
|
|
03-21-2016, 02:48 PM
|
#31
|
Human being with feelings
Join Date: Apr 2014
Posts: 943
|
Yes Justin within Source Beat mode editing, this will never happen.
Because the project timeline is not involved...
It also never or say seldom happen, if you resize the left edge of MIDI item with snap to 1/4
and of course move item with snap, use snap markers without snap and moving after
with snap again etc pp.. puuh
And it normally always happen, if you resize left edges once and or move items once without snap. and do editing after this..
Personally for me it is no problem. I never resize or move MIDI Items without snap and for light offsets I use the snap-marker,
but many other User do it not this way.
I don´t believe, that this is only one display issue..
Just use for example 120ppq with 30 BPM, 1 Tick would be ~16,667 ms
And the result is 16,66 ms difference
The first MIDI item include the 1 tick "display" issue with 120ppq at 30 BPM
and it has trigger the first resulting recorded audio from synth(track 3)
The second MIDI item has no 1 tick "display" issue and the resulting audio is at
track 4..
You can clearly see the difference.
With 1 tick it is 17ms (120ppq, 30 BPM)
If it is only an display issue, there would be no difference^^
For me something is not in sync with the project timeline beat display, item position vs Midi item editing or whatever it is and for me that is also the reason why Note off release values at exactly loop end never transmitted within loop.
Only after you shift these note offs at least 1 tick before loop end, otherwise only 0 for release is transmitted...
If I now think about these 1 tick thing....
################################################## ######
EDIT: 22.03.16
FIXED with 5.17rc1
incl. Note Off release at loop end position
################################################## ######
__________________
I hope you can understand me? Without german beer my written english is always very bad, with beer it becomes unbearable!.
Less is more! To much limited the own creativity.
Last edited by ELP; 03-22-2016 at 01:06 AM.
|
|
|
03-21-2016, 07:15 PM
|
#32
|
Human being with feelings
Join Date: Dec 2014
Posts: 543
|
Here I copied a midi item twice using the paste as pool action. The first paste is fine. The second paste is off.
If I delete the third copy, and then do a new copy Ctrl-c on the original, and then paste it again, it was still off. But if I move over one measure, and paste it there, it's ok. But if I then copy that just-pasted one, and paste it right next to it, that one is off.
Last edited by panicaftermath; 03-21-2016 at 07:22 PM.
|
|
|
03-21-2016, 07:47 PM
|
#33
|
Human being with feelings
Join Date: May 2012
Location: Brighton
Posts: 1,989
|
Well, then I'm not hallucinating those things. When I was coming from FL to Reaper, I did notice those issues almost instantly, but not being accustomed with the technical side of things ( didn't even know there is such thing as events list/editor until recently), as well as having no one around who could confirm those observations, I was left out in the dark thinking thy there must be something wrong with me if everyone says that everything works fine. But those and other issues where present in v4 too ( referring to OP as well as post above and some of my own threads ).
|
|
|
03-24-2016, 06:05 AM
|
#34
|
Human being with feelings
Join Date: Jun 2015
Posts: 8,704
|
I normally play the music in with MIDI keyboards but I just tested this to confirm the issue and it happens for me too (with current build Reaper).
|
|
|
03-24-2016, 06:33 AM
|
#35
|
Human being with feelings
Join Date: Aug 2007
Location: Near Cambridge UK and Near Questembert, France
Posts: 22,754
|
.041 in ticks....
OK next question is: does this get more and more "out" with multiple pastes?
FWIW I cant reproduce this so far....
__________________
Ici on parles Franglais
|
|
|
03-24-2016, 06:53 AM
|
#36
|
Human being with feelings
Join Date: Jun 2015
Posts: 8,704
|
Quote:
Originally Posted by ivansc
.041 in ticks....
OK next question is: does this get more and more "out" with multiple pastes?
FWIW I cant reproduce this so far....
|
When you change your figure if it works fine try upping the figure to the next decimal place figure. I tried 3.1.75 to 3.1.77 it immediately drops to 3.1.76.
|
|
|
03-24-2016, 10:59 AM
|
#37
|
Human being with feelings
Join Date: Jun 2015
Posts: 8,704
|
For what it's worth I tested this in the current builds of Sonar Artist, Mulab demo and Cubase 8 demo. None of those had this issue.
Still happening in 5.16 too.
|
|
|
03-24-2016, 01:49 PM
|
#38
|
Human being with feelings
Join Date: Apr 2014
Posts: 943
|
It is fixed with V5.17
__________________
I hope you can understand me? Without german beer my written english is always very bad, with beer it becomes unbearable!.
Less is more! To much limited the own creativity.
|
|
|
03-31-2016, 09:43 PM
|
#39
|
Human being with feelings
Join Date: Dec 2014
Posts: 543
|
This did seem to be fixed. But I'm seeing this in 5.17.
|
|
|
03-31-2016, 10:08 PM
|
#40
|
Human being with feelings
Join Date: Dec 2014
Posts: 543
|
Check out what's happening with the very first event, which starts out quantized on x.1.000 and when the strength slider is moved to 100% ends up at x.4.959.
Last edited by panicaftermath; 03-31-2016 at 10:15 PM.
|
|
|
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 02:16 AM.
|