Custom shape end point smoothing. Always avoid pulling handle?

[Annabanna]

Experiments I've done so far with the custom shape window in single trigger Modulator LFO's seem to indicate that the smoothing handle of the endpoints shouldn't be pulled as doing so smooths beyond the point itself which, in the case of endpoints results in the curve being abruptly squared off to reach it's mid-value at the end of the cycle.





Is there some other circumstance in which pulling the handle is advantageous? (aids in smoothing looped LFO's, or when used in some other mode, for instance)... or am I missing some control or setting that affects this even under the circumstances I've tested?

[MeldaProduction]

I'm afraid I don't really understand your question, again. A language barrier or something, after failing at so many of your questions I'm starting to feel dumb really...
Anyways I'm guessing you may be asking why sometimes the smoothing cannot really be set (or set "big"). That's because the engine makes it small enough so that it can actually provide the requested smoothing without some brutal artifacts, like super high peak in between.

[Annabanna]

Some of these things simply are not easy to describe verbally. I understand that.

That's why I included pictures. Look at the bottom one. See how it jumps up about 2/3 of the way to the right? That is the result of the smoothing handle being pulled out in the pic above.

If you push the smoothing handle in, the jump in output disappears, but the overall smoothing is more abrupt leading up to it (as is always the case when pushing in smoothing handles). I'm just wondering if that means I should ALWAYS keep the handles in for endpoints in single trigger mode, or if there is some setting that smooths out the last point since engaging the smoothing handle necessarily extends the curve beyond the window frame.

[MeldaProduction]

I'm afraid I'm lost again - for instance I have absolutely no idea what is on the second picture... ???

[Annabanna]

I'm afraid I'm lost again - for instance I have absolutely no idea what is on the second picture... ???

First pic is the shaape the user selects. Second is what that shape looks like on output. Look at the contours. They're the same. The jump in the second pic corresponds exactly to the end of the frame on the right of the top pic. It happens to be a waveform display in this case as the modulator happens to be controlling amplitude, but it doesn't matter because it's just showing the modulator's behavior. If the modulator were modulating any other parameter, a picture of that parameter's output would look the same.

It's OK. I've done enough experiments, and measured enough outputs at this point, I don't really think there's an open question any more so much as a conclusion:

Don't ever pull the smoothing handles of endpoints... at least not in single-shot mode.

[MeldaProduction]

Hmmm, I'm still lost, are we even looking at the same image??? The second image here looks like some sort of distorted waveform with some cyan crossing of no purpose . Anyways if it is ok, then it is ok!
Btw. the shape an envelope editor shows should be the exact shape it creates (if that was somehow in the question).

[marzelli]

Btw. the shape an envelope editor shows should be the exact shape it creates (if that was somehow in the question).

The sudden jump in the middle of the waveforms envelope which obviously shouldn't be there. Even I can see it

[Annabanna]

Hmmm, I'm still lost, are we even looking at the same image??? The second image here looks like some sort of distorted waveform with some cyan crossing of no purpose . Anyways if it is ok, then it is ok!
Btw. the shape an envelope editor shows should be the exact shape it creates (if that was somehow in the question).

Yes. There is a cyan crossing of no purpose. If I had thought ahead, I might have lined it up with the end of the envelope frame in the first pic so I could show the correlation. It's really only the shapes that matter. None of the other details are important.

It's not really a problem. I just noticed a few cases where pulling the handle extends smoothing beyond the frame that is visible on output, but not accounted for in the editor image, so I was wondering if it might be the result of some setting. It doesn't appear to be, so just something for me to keep in mind moving forward. It makes sense in general that smoothing past the envelope frame might be an issue. I was just confused because it didn't match the editor image or preview.

It's a similar issue, in a way, to how Logic has an option when bouncing as to whether or not you want to include reverb trails. All the notes may be done, but that doesn't mean nothing extends beyond the endpoint.

Actually, such an option would be quite helpful here... "Extend smoothing beyond envelope frame." I'll get you a better image to demonstrate.

Thx for the response.

[MeldaProduction]

marzelli: There is sort of a jump in the waveform, sure, but it looks nothing like the first image and the "random peaks" just don't have much useful information.

annabanna: It doesn't really works that way I'm afraid. There are no frames or anything... It's important to understand that smoothly connecting 2 segments with a different (possibly changing!) gradient is hard and sometimes impossible (at least without a jump out of range).

[Annabanna]

OK, I've redone this to make the shapes more obvious. To clarify, the "random peaks" have nothing to do with anything. If I had limited the white noise on the way in, it would just be a solid line. Again, it is only the shape that matters. The editor clearly shows no jump. The output clearly shows a jump.

Here, no random peaks:

Doing this:



Produces this:



Note that the smoothing handle for the endpoint on the right of the editor window is not pulled. So far, so good.

Now, pulling the smoothing handle looks like this in the editor:



And produces this:



In this case it's clear what's happening. You can see in the editor that pulling the handle extends the smoothing beyond the editor frame. As a result, the curve jumps abruptly on output because it's not done smoothing by the time the single triggered modulation ends.



What I was pointing out in the OP is that I've been noticing cases where this jump happens even though the editor and preview windows give no indication. (take another look at the OP pics in this context, and this should be fairly clear. Again, the individual peaks are meaningless. I could have just as easily forced it to be a solid shape. It just didn't seem necessary as you can just look at the overall shape)



FR:

It might be more productive to take a step back, though. There's an opportunity here. If there were an option to "Extend smoothing beyond frame" which kept the single trigger going until the smoothing had finished, then behavior would match user intent in all cases, and it would allow for much better smoothing options to be used without issue. The "Include Audio Tail" option I mentioned when bouncing in Logic addresses a similar issue.

I understand and respect that you do not want to complicate anything, but I would suggest that this simply allows it to do the thing any user would necessarily be intending anyway if they pulled the handle. Common sense would indicate that if they didn't want to extend the smoothing, they wouldn't have pulled the handle. In cases like that in the OP where the jump in output occurs even though there is no indication in the editor, I'd say it's a safe bet that most users are expecting full smoothing, and simply not aware that it is cutting off instead. Either case would just fix itself if there were an option to extend the triggered modulation until the smoothing had completed.

[MeldaProduction]

I'm afraid that won't happen, sorry. It's technically far more complicated than it looks and I just don't have the development time for such a superadvanced thing, especially since nobody would probably notice it .