[Ardour-Users] Location point

[Lamar Owen]

On Wednesday 21 April 2010 01:16:06 pm Thomas Vecchione wrote:
> That would make sense to me.  I would want to find the cause of the
> automation points of course.

[I know this isn't a Mixbus support list, but I've seen some issues with 
several versions of Ardour as well as Mixbus that seem to be related to this, 
so I'll post my observations here.....]

Automation points are and have been weird, in my usage, and in a way that has 
made it very difficult, at least for me, to gather the information needed for a 
useful bug report.  And I prefer working in that visual mode rather than open 
the fader automation waveform and doing it that way, although I realize that's 
backwards to the way most work.  I guess it's just bad habits learned from CE, 
CE2k, CEP, Audition, and other similar editors, over the past two decades....

If you drag an automation point around while the playhead is moving, and the 
canvas scrolls, the automation point ends up in wild places, for instance.  
Over the past year, I have built sessions that have unmoveable and undeletable 
gain points in portions of tracks that have no audio, even (when I say 
undeletable, it's because deleting the point resulted in a complete crash; 
I've not attempt to delete it since).  I would love a quick way to clear out 
all automatic gain points on a track in one fell swoop.  I realize I could 
probably hand edit the session file to do so, but just haven't committed the 
cycles to learn how to do that yet.

Now, I'm primarily using Mixbus on OS X at the moment, but I have continual 
issues with automation points, even with 1.2.  In fact, I have more issues 
with 1.2 than with 1.1 under Mixbus.  I haven't tried lately with vanilla 
Ardour on Linux or OS X, however.  I can reproduce very easily, and will be 
building up a session file (that's not client-rights-encumbered) with the issue 
as soon as I can get the time to do so, for Harrison's debugging benefit.  Of 
course, there has been a flurry of SVN traffic in the mixbus branch as of late, 
so it may be fixed for all that I know.  But I know that in doing a simple (and 
I mean simple) three region radio broadcast, I can reproduce the issue I see 
every time, using MB 1.2.

With automation gain points in a simple arrangement (a radio broadcast 'intro' 
with a fadeout, overlapping the actual broadcast content for thirty seconds or 
so, then the 'lockout' fading up for 45 seconds or so under the content, with 
all fades handled by envelope automation gain points drawn directly on the 
waveforms (not in the automation drawing line)) I can export, play, etc 
anywhere in the timeline without issue as long as I don't exit Mixbus.  

Once I exit, and reopen the session, I get an ERROR at the bottom talking 
about not being able to reconstitute a Memento or something like that (I know, 
I know, I need to get the exact text, but I haven't had time yet, as when I 
have used MB lately it's been under time pressure, and as long as I know the 
workaround (delete the intro region, re-import the intro, re-make the gain 
points, as there are only four to make) I can do the workaround faster than I 
can properly build a bug report....sorry....).  

After reopening the session, if I put the playhead on the timeline after the 
content starts (a minute or so after the intro starts) it plays; but if I put 
the playhead to a point prior to the content start, I get the 'clicks of 
death' meaning the LADSPA plugin connections have gone bad (there are no 
plugins loaded other than the Harrions channel strip/mix engine one).  So I 
use the workaround of putting the playhead after the start of the content, 
removing the intro region, re-importing the intro, building the four-point 
fadeout, and I can work again.

This sort of gain point randomness has been an aggravation to me for a year or 
so, during which time I've used Ardour 2.7.x and 2.8.x, under several Linux 
distributions, and Mixbus under OS X.  I've had gain points disappear 
completely, I've had gain points go to impossible gain values (plus 120dB of 
gain in one instance), I've had gain points go to impossible places on the 
timeline, and other things.  Mixbus 1.1 was fairly stable, but 1.2 has been 
annoying.  Most of the issues occurred when I let the timeline/canvas scroll 
while dragging a point, if that's even a helpful data point; I no longer move 
or create gain points while playing.

I've even watched gain points disappear; if I want to bring up the gain on a 
three or four note guitar run, for instance, instead of using fader automation 
I've always used the envelope gain points, raise the gain a dB or three for 
the run, zoom out, save changes, then exit.  When I reload the session, I get 
loads of those Memento errors, and I've lost all but one of the four gain 
points used to draw in the gain increase on the run, which means I have to 
rebuild them all.  I've also used gain points to squelch out accidental peaks; 
one of my clients plays five string banjo, and tends to occassionally hit the 
fifth string very hard; three gain points to draw a sharp dip of 10dB or so on 
top of the twang and it's fixed; at least until one of those gain points just 
up and disappears.

I've found that if I don't zoom out to where the four (or three) points look 
like one, that the points tend to stay put, at least on Ardour 2.8.x, where I 
used gain points a lot more than I use them with Mixbus (emulating a leveling 
amp; MB has that function on the channel strip, so I don't use gain points for 
that anymore).  

But I've fought with this for a long time.  But it's never been to where I 
could file a proper bug report; mostly because the corruption has been random, 
and has been seemingly dependent upon file open order (I have one session that 
causes the 'click of death' LADSPA disconnects only if its the first session 
loaded; if I load another session first, then open this session, it works fine, 
although the last time I opened that session was on MB 1.1), making it very 
difficult to get enough information to file a useful bug report.  And I'd rather 
not file a report than file a useless one.  

So I keep trying to get the information I need to file a really useful report, 
but thus far I've not been successful.  And just to be clear, I'm talking 
about gain points drawn directly on the audio waveform, not on the automation 
waveforms for fader, panner, or plugin control automation.

And while it has been a while, I've seen the end point go way way out in the 
timeline before, too.  But it's been a while, and I don't remember what 
version that was, sorry.  These issues have just been so random that 
reproducing and documenting them for a good bug report has been a real chore.