Another post about setting slope values

[UpTilDawn]

While I'm at it, I ran into a very odd thing tonight while automating multiple tracks to fade out. I wanted to use a single fader entry on each track, all set at the same starting point (having grouped the tracks first in a mixer view). I then wanted to adjust the fader slope value on all these tracks (selected as a group in the MT).

The fader slope value was an easy setting to make and all selected tracks appeared to make the same adjustment. However, on playback not all the tracks faded at the same rate.

At first, I discovered that not all tracks faded to infinity with this single adjustment, so I made the necessary tweaks and re-automated. Even then, they would not all fade out at the same end point. One track, in particular, was stubborn as can be and would not fade out until very near the end of its region (which went beyond the actual end of the sound on that track). All selected tracks had the same region length and all were set to fade out with the same slope value and end at infinity, yet all fades did not end by the end of the slope.

I made allowances for some tracks to end early, expecting that this would add a little realism to the overall sound. I cannot explain why one track would go far beyond the end of the fader mark. Adjusting other tracks individually created simliar odd behavior, but nothing I paid attention to at the time. This one track would even fade out after all the others when I adjusted its slope value independent of the others and even after sliding the fader mark forward a couple of times.

I finally found a slope setting that allowed it to fade out before the end of its natural sound and then adjusted its fader mark backward until the sound was balanced with the other tracks..... but I still don't know what's wrong.

I was using the SAW Levelizer as a post FX on three or four of these tracks to contain the peaks and this is the only thing that strikes me as a plausable explaination for this behavior since the limiter causes Saw to adjust buffers.

DanT

[Bob L]

If a fader entry is set at the same point on multiple tracks with the same slope value... they will all fade at the same time... but, the audio may sound different and certain tracks may linger in the mix longer as they fade because each audio signal will react differently to the same fade steps depending on what that signal is.... but you must make sure all slopes finish before the end of the regions... if any extend further, then that would cause some maybe not to complete to infinity.

I did notice that the slope display may visually end a little short of the actual final position in some cases... I will look into that.

For better results easier... I would suggest you simply cut each track's final region and use a softedge of varying degrees on each one... then you can easily adjust the ramps of each sudio signal to suit what you want to hear and the final fadeout sounds extremely natural.

Bob L

[Cary B. Cornett]

... I would suggest you simply cut each track's final region and use a softedge of varying degrees on each one... then you can easily adjust the ramps of each sudio signal to suit what you want to hear and the final fadeout sounds extremely natural.

Bob LI have done exactly as Bob suggests and found it to work very well. In fact, whenever I want a linear fade out or fade in, I now prefer using a SoftEdge to do it, rather than writing fade automation. It is quicker, easier, just STUPIDLY easy to readjust length and position of fade until I get what I want, and writes a smaller EDL (probably runs more efficiently too...).

What is that? A win win win win... ;)

[Dave Labrecque]

At first, I discovered that not all tracks faded to infinity with this single adjustment, so I made the necessary tweaks and re-automated. Even then, they would not all fade out at the same end point.

Bob, it sounds like he's saying the fader entries were all -inf, and so SAW should have faded out all the way on all tracks by the end of the sloped entries, no?

Dan, is this correct? Are you saying that some tracks did not fade all the way out when they should have?

[Bob L]

Dave... but if some of the slopes ended after the region end... this would cause those to possibly stick.

I just tried it here and used the Levelizer on each track also set to peak limit and each track faded the same and to complete infinity when listening in my headphones.

VST latency causing plugins on tracks could also be his problem... they would shift everything out of sync.

Bob L

[Dave Labrecque]

I have done exactly as Bob suggests and found it to work very well. In fact, whenever I want a linear fade out or fade in, I now prefer using a SoftEdge to do it, rather than writing fade automation. It is quicker, easier, just STUPIDLY easy to readjust length and position of fade until I get what I want, and writes a smaller EDL (probably runs more efficiently too...).

What is that? A win win win win... ;)

I agree with all you've said. I only wish there was another fade shape option for SoftEdge. So often when using it for these end-of-region fade-outs, it just doesn't flare the end as I'd like, so the last bit of audio ends a bit abruptly. That's when I reach for the standard fade tools. Which aren't so heinous. :)

[Bob L]

Adjust the region backward and use a longer softedge... works everytime for me.

Bob L

[Dave Labrecque]

Adjust the region backeward and use a longer softedge... works everytime for me.

Bob L

Interesting, that hasn't always worked so well for me. In my experience if the audio content cooperates (quieter audio at the relocated -inf point of the fade) it can work , but often I'm not so lucky. :)

[Dave Labrecque]

Dave... but if some of the slopes ended after the region end... this would cause those to possibly stick.

I just tried it here and used the Levelizer on each track also set to peak limit and each track faded the same and to complete infinity when listening in my headphones.

VST latency causing plugins on tracks could also be his problem... they would shift everything out of sync.

Bob L

Might be a good idea to code around that "sticky" possibility, eh? I'd think audio should end at the end of a region (or it's SoftEdge), regardless of what the fader automation is doing, no?

VST latency-causing plugins? No way to code for those issues, I imagine. :)

Question: I understand it's not flawless (so many plug-in developers in the universe), but if the VST latency compensation is being handled successfully by SAW, are such things still an issue? Should we expect some automation to be out of sync with region data for all instances of latency-inducing VST plugins?

I just want to be sure my expectations are in-line with reality as I make my editing/mixing choices.

[UpTilDawn]

Let's see if I can get this all in one post without re-reading......
Thanks for the quick ... ultra-quick replies guys.

Yes, softedge works gret a lot of the time, but sometimes I too want a different kind of fade at the end, especially on a long instrumental fade out.


All region boundaries extend well beyond the end point of the fader slope.

No VSTs... basic channel strip settings, minimal automation, three or four Levelizers (including this track) set to post fade positions, no channel strip compression or gate... only volume and eq settings.

Yes, what I am saying in my long-winded style :D is that all channels did not fade out equally by the end of the fader slope, although all but one faded out by the end of the slope..... One track faded well beyond the end of the slope mark, but before the end of the region boundary (which extended well beyond the end of the sound on the track).

There's something wrong that I can't quite get a handle on. I'm still puzzled by what caused this. As an example, I originally set six tracks to fade out before the actual ending of the sound using a single automation entry that adjusted the group of tracks more or less to infinity... Some of the tracks did not initially go to inf., so I further adjusted them later.

I adjusted the slope value for all six tracks to 20,000ms. All but one track faded out by the end of the fader slope marking. One track continued to fade slowly and was still audible when its sound ended in the original recorded track, which was well within the region boundary.

I then played around with the slope value of the one track until I could get it to fade out before the end of the track. This was at 15,000ms.... and too short of a fade. I then tried adjusting it upward to get a fade time that matched the others, but when I got to 17,000ms it faded as though it was set to 20,000......

I don't get it.:confused:

[Dave Labrecque]

Let's see if I can get this all in one post without re-reading......
Thanks for the quick ... ultra-quick replies guys.

Yes, softedge works gret a lot of the time, but sometimes I too want a different kind of fade at the end, especially on a long instrumental fade out.


All region boundaries extend well beyond the end point of the fader slope.

No VSTs... basic channel strip settings, minimal automation, three or four Levelizers (including this track) set to post fade positions, no channel strip compression or gate... only volume and eq settings.

Yes, what I am saying in my long-winded style :D is that all channels did not fade out equally by the end of the fader slope, although all but one faded out by the end of the slope..... One track faded well beyond the end of the slope mark, but before the end of the region boundary (which extended well beyond the end of the sound on the track).

There's something wrong that I can't quite get a handle on. I'm still puzzled by what caused this. As an example, I originally set six tracks to fade out before the actual ending of the sound using a single automation entry that adjusted the group of tracks more or less to infinity... Some of the tracks did not initially go to inf., so I further adjusted them later.

I adjusted the slope value for all six tracks to 20,000ms. All but one track faded out by the end of the fader slope marking. One track continued to fade slowly and was still audible when its sound ended in the original recorded track, which was well within the region boundary.

I then played around with the slope value of the one track until I could get it to fade out before the end of the track. This was at 15,000ms.... and too short of a fade. I then tried adjusting it upward to get a fade time that matched the others, but when I got to 17,000ms it faded as though it was set to 20,000......

I don't get it.:confused:

Good news... either you're doing something wrong or SAW's doing something wrong. :) You're not crazy. If you send me the EDL, I'll have a look. I'm sending you a PM.

One other thing to check: are you certain the problematic sloped automation entry is set to -inf dB?

[Ian Alexander]

I did notice that the slope display may visually end a little short of the actual final position in some cases... I will look into that.
Bob L
Loving the visual aspect of this approach (marking an area, noting its length, using that to set the slope), I have noticed that the visual length of the slope often needs to be tweaked quite a bit to make the fade actually run the desired length. So, this will be a very helpful fix.

Whatever the slope, it does tend to drop off a cliff at the very end, though.

I will try the softedge approach, too.

[UpTilDawn]

I've used both the fader slope and softedge at the same time to good effect when the fader slope (or the softedge for that matter) falls off too quickly at the tail end of the fade.

danT

[Dave Labrecque]

Whatever the slope, it does tend to drop off a cliff at the very end, though.

I will try the softedge approach, too.

I think SoftEdge suffers the same "drop off a cliff" issue. It's in the fade-type Bob's hard-wired in. My feeling is a selectable curve type would be a great addition to SoftEdge, perhaps to sloped entries, too, especially in these single-entry, long-slope-time instances.

I think it's fairly common for DAWs to offer at least a couple fade shapes for automatic crossfades, etc. I'm sure Mark will let us know. :p

[Dave Labrecque]

I've used both the fader slope and softedge at the same time to good effect when the fader slope (or the softedge for that matter) falls off too quickly at the tail end of the fade.

danT

Interesting. Kind of a work-around, but very innovative approach using the tools at-hand. :)

[UpTilDawn]

Good news... either you're doing something wrong or SAW's doing something wrong. :) You're not crazy. If you send me the EDL, I'll have a look. I'm sending you a PM.

One other thing to check: are you certain the problematic sloped automation entry is set to -inf dB?

Well, I've some interesting news to report thanks to Dave's offer to look at the edl.
Here's the note I just sent to Dave in an email. I thought maybe it might be useful to track down this bugger.

Well Dave.

You inspired me to try a couple of things.

I opened a copy of the "problem extract" edl I sent to you and inserted a short song file to the problem track..... which is track 8, by the way.

I copied the automation to this new soundfile and the same problem existed.

I tried replacing the automation with one from a track that had faded properly and the same problem still existed.

Lastly, I moved the problem soundfile with its automation to an adjacent track and the problem no longer exists.



Strange, but I seem to remember this moving a region to another track as being the solution to other problems mentioned on the forum before.

Don't know what the bug is..... Do you?

I think I'll post this email to the message thread and see what others (Bob) says about it.

Thanks for your assistance.

DanT

Also, the fades actually end beyond the visible end point of the fade automation mark in all the tracks, but not far out like the problem track.
The problem appears to be tied to the track and not the fade mark or the region....... which is pretty strange to me, but as I mentioned to Dave, I remember this kind of issue cropping up in past posts.

Hope somebody can figure this out from what I've provided........ BOB:D

DanT

[UpTilDawn]

I also did an extract session edl in which I only took the 40 or so seconds from just before the automation in question to beyond the end of the fade and discovered that I do not need the original soundfiles to duplicate the problem.

I think it's safe to say that if anybody wants for me to send the edl, you could insert any soundfile in place of the blank regions that are created when you open the edl and use ctrl/ok in response to the request to find the files. You can then hear for yourselves what the problem I am experiencing here is all about and try moving and copying things around to see how the problem goes away when you move the region and automation to a new track.

DanT

[Bob L]

I did check out the slope display length being different from the actual fade time and here is what I found:

The calculation results for the level factor changes over the length of time during the slope are stored and applied in integer and are rounded down... so a ms slope length that gives a value change of 1.585 in level factor for each sample, for example, would round down to 1... this then causes the fade to take slightly longer than if the decimal values were carried throughout the factor adjustments during the slope.

The slope values for each automation entry are most often are very small, such as a few ms or few hundred ms and therefore this slight roundoff difference is usually insignificant. With large ms values, creating slope fades of 10 or more seconds, the roundoff can cause the fade to take noticably longer than the actual slope length displayed (which is accurate), depending on the resulting decimal result and amount of roundoff.

The performance advantages of the integer adjustments and the non-critical nature of the resulting time discrepency on very large slope entries, suggests to me that I leave the code as is for now.

Regardless of the time discrepency in how long the fade takes, the actual math that does the fade still should result in a perfect and complete fade to the destination value.

I have not been able to duplicate the stated issue of a slope not completing to a finished fade value...

As to the idea of a fade dropping drastically mid-fade... the only thing I can imagine that would cause that is if individual entries in the middle of the fade have slope values that extend past the next entry... when a slope value is interrupted, the results can appear random.

I would highly recommend that long fadeouts like 10 and 15 seconds, use the actual F command fade and then the entries be adjusted to plateau near the end to taylor the final drop off to suit your desires for the individual audio situation.

Bob L

[UpTilDawn]

Thanks a ton for looking into this Bob.

The fade extending beyond the visible mark doesn't bother me all that much... except in the one case I had trouble with yesterday.

When it gets to be problematic, I'll resort to a stepped sequence of fader marks and slope them as you suggest.

The mid-fade drop situation contained just the one fader mark and a somewhat long slope value. This is on the output track (I'm fried right now, so I'm not sure I remember correctly.) and is consistent in its behavior.

Again... THANKS for checking it out.

DanT

[Carl G.]

Loving the visual aspect of this approach (marking an area, noting its length, using that to set the slope), I have noticed that the visual length of the slope often needs to be tweaked quite a bit to make the fade actually run the desired length. So, this will be a very helpful fix.

Whatever the slope, it does tend to drop off a cliff at the very end, though.

I will try the softedge approach, too.

I found Softedge as a far superior way to get quite fades with no dropping off the cliff. That sold me.
It was impossible to get the a smooth but rapid natural fade through complex using the fader in any slope form (before the next large snare hit).

However.... I'm still getting used to adjusting fades with Softedge....
because I lived with pots and Faders ALL my life. :)

[Ian Alexander]

I found Softedge as a far superior way to get quite fades with no dropping off the cliff. That sold me.
It was impossible to get the a smooth but rapid natural fade through complex using the fader in any slope form (before the next large snare hit).

However.... I'm still getting used to adjusting fades with Softedge....
because I lived with pots and Faders ALL my life. :)
Just occurred to me that SoftEdge also has an advantage if you have any regions following the fade. You don't have to fade up again. Hmmm... I may have to try it for this situation.

[UpTilDawn]

I found Softedge as a far superior way to get quite fades with no dropping off the cliff. That sold me.
It was impossible to get the a smooth but rapid natural fade through complex using the fader in any slope form (before the next large snare hit).

However.... I'm still getting used to adjusting fades with Softedge....
because I lived with pots and Faders ALL my life. :)

On quick fades, softedge is great...... On long fades it sometimes leaves me wanting more... although I really like using softedge for most fades these days.

You really have to watch yourself when it comes to softedge and reverse audio regions. I never get what I expect at all without lots of trial and error.

DanT

[Dave Labrecque]

Bob, Dan,

Here's what I've found...

The lower the starting gain value for the track, the shorter the executed sloped-fade-time is. And the higher, the longer.

I looked at Dan's "extracted" EDL. Though all the automation entries are identical across the tracks -- all dropping to -inf over 20000 ms -- they all fade out at different times. It appears that the gain value prior to the automation entry determines when the audible fade finishes, and dramatically so. 1 dB of starting gain difference translated to over two seconds of difference in the fade-out time (in the vicinity of initial unity gain, anyway).

My guess is that the generated fade curve is identical in each case, only offset by the gain amount, and so the intersection of the fade curve and the -inf lower limit is at a different part of the curve in each case when the cursor gets to -inf, and therefore at a different point in time. If ya know what I mean.

I think the expectation is that the fade will be rescaled between the starting gain and -inf in each case, rather than simply offset.

Bob?

[Bob L]

Dave... its a simple calculation... it was never meant to recreate any accurate fade curve... it is simply a way to ramp the signal down faster or slower.

The calculation takes the difference between the start and end levels and does the math to come up with a delta value to drop for each sample in the loop to finish by the slope length time... the final results will vary do to the fact that the result is stored and used as an integer value... so if the actual delata calculation is 2.568 it will be rounded down to 2... if its 2.999 it is still rounded down to 2... so there will be different discrepencies based on the actual values and slope lengths.

Again... it is simply an estimated drop over time.

Bob L

[Dave Labrecque]

Dave... its a simple calculation... it was never meant to recreate any accurate fade curve... it is simply a way to ramp the signal down faster or slower.

The calculation takes the difference between the start and end levels and does the math to come up with a delta value to drop for each sample in the loop to finish by the slope length time... the final results will vary do to the fact that the result is stored and used as an integer value... so if the actual delata calculation is 2.568 it will be rounded down to 2... if its 2.999 it is still rounded down to 2... so there will be different discrepencies based on the actual values and slope lengths.

Again... it is simply an estimated drop over time.

Bob L

OK, I think I get it now, thanks. I guess the bottom line is that you're okay with varying fade times (depending on the delta factor) for a set slope time -- most noticeable only on the longer slope times, of course. I guess I wasn't expecting that. :)

I guess the lesson for Dan is that in SAW that's not the way to do what he wants to do.

[UpTilDawn]

Hi guys.
Thanks for working this out for me.:)

After doing a few more test fades myself, I found out a few ways long slope values across multiple tracks can still be worked to my advantage. I just have to be able to expect that the result I hear won't equal what I see on the screen.

I can actually put this to use in a way similar to Bob's use of softedges across the ends of multiple tracks to create a natural sounding fade out, where each instrument fades at slightly different times/pace.

It's nice once you learn what to expect.

DanT