Support and General Use > Audio Playback, Database and Playlists

Crossfade settings

<< < (2/2)

Falco98:
Ok, I've started drawing it out on paper and it's starting to make some sense to me.

The reason for the 2 different "delay" settings might be so that someone can do one or another desired effect (well duh).. but it's more flexible than just one "overlap" setting, as far as I can tell.

By default, I assume (and would like some solid input here from a dev), Fade-out and Fade-in both start at exactly the same time.  That is, whatever "duration" you've set for each, the second the old song starts fading away, the new one starts fading in.  I'll assume this is true just for now...

So let's say you want a nice long (7s) fadeout on song A.  But you also want a quick fadein for song B underneath that (let's say 1s).  The problem is, the last 6 seconds of song A will be heard underneath a full-volume song B, which only took 1 second to fade in.

So let's use fade-in delay to help us out.  Let's say we want the old song (A) to be nearly done by the time the new one (B) comes in.  If we set the fade-in delay to 5, then A will fade out for 5 seconds before B fades in (starts?), and the overlap will be 2 seconds.  That seems pretty reasonable (again i need confirmation from a developer).

Another scenario:  We want B to fade in slowly underneath A, and then A to suddenly drop out.  But if we set fade-in duration to 7 and fade-out druation to 1, the effect (devs?) will be that A suddenly fades out while we're still waiting around for B to slowly fade in.  So we set the fade-out delay to, let's say, 5 again:  now B will slowly pick up underneath A for ~5 seconds, wherein A will drop out just before B reaches full volume.  That seems pretty acceptable.

The one thing I think others are right about is, if both delays are set to the same value, they cancel each other out.  I'd even go so far as to guess that if one is 4 and the other is 2, they may as well be 2 and 0... but i'm not sure.  Again it'd be nice if a dev could lend some clarification.

Mmmm:
Yep...I think you've pretty much got it there!  :)

Chill:
Falco98

There's nothing like a diagram, or some examples like that, to make a clear explanation - thanks.

I still maintain that crossfade only requires 3 parameters to fully describe it, and therefore that one of the four we have now is redundant.  All we need to specify is how long song A takes to fade out, how long song B takes to fade in, and how much they should overlap.

So in your scenario A, we could uniquely describe the desired effect with:
Fade Out Duration = 7s
Fade In Duration = 1s
Overlap = 2s

Equally, scenario B could be uniquely described with:
Fade Out Duration = 1s
Fade In Duration = 7s
Overlap = 6s

I can't think of a scenario where the intended effect could not be specified with these three parameters - anyone?

Personally, even though I understand the current options now, I still think they are not intuitive - the option of 4 parameters leads to confusion, like your final example for instance.  My own preference would be for a three parameter setup in the GUI.  I can well imagine that the 4 parameters might be useful/necessary in the crossfade code, but if that is the case, then there must be a simple algorithm that could calculate Fade In Delay and Fade Out Delay given these three parameters.  I know - if I'm so keen on this, why don't I write a patch?!  Well that's another good question...

EDIT:
Actually, the relationship between Fade Out Delay (call it FODe) and Fade In Delay (FIDe) is simple:
FODe = -1 * FIDe.  If one is negative, multiply it by -1 and call it the other one.

The algorithm to calculate them from the Overlap (OL), Fade In Duration (FIDu) and Fade Out Duration (FODu) is also simple (in fact, only the OL and *EITHER* FIDu or FODu are needed to calculate both of the delays).  In terms of OL and FODu, the algorithms for FIDe and FODe are:
FIDe = FODu - OL
and therefore FODe = OL - FODu

I've tried this on both of your examples Falco98, as well as a couple of other examples: one where there is a silent gap between the tracks (OL = -ve) and one where the OL is greater than the sum of the durations (ie the both tracks are at full volume for a period).  It works in all cases (until someone shows me the case that I've overlooked of course!)

Falco98:
I suppose you're right, though just having a "negative" overlap might be confusing to users (and may be very difficult to program, but also maybe not).  But all that would do in the program is convert back to the FIDe and FODe values, so from the programmers' point of view, there's no real good reason to do so :-P

hhannah:
Man all this REALLY did help me out! THANKS!!! Falco98's post (Posted on: May 05, 2006, 09:11:51 AM) really puts it clear. The manual still doesn't explain much, and the subject is not trivial at all, so it would be great if somebody would put a nice explanation like this on it, and (I totally agree Chill)... A diagram and a few examples... please!

Navigation

[0] Message Index

[*] Previous page