Exactly. FF1 is from Compressor 1, FF2 is Compressor 2, both from earlier versions of Ableton. FB was new in v7 or v8 I think, but you can't use FB with a sidechain because the control signal doesn't appear at the output.jobbanaught wrote:Dare to explain FF1 and FF2 to me? Afaik FF means feed-forward in contrast to feed-backward, but i dont know the exact difference. Nor do i know the difference between FF1 and FF2.damagedgoods wrote:Have you tried the two different compression modes (FF1 and FF2), and also using the filter on the sidechain (especially good for kicks - if you highpass them they decay a lot quicker)? Makes a big difference.
ableton compressor (sidechain)
-
- mnml mmbr
- Posts: 349
- Joined: Tue Feb 12, 2008 1:38 am
-
- mnml mmbr
- Posts: 349
- Joined: Tue Feb 12, 2008 1:38 am
The difference between FF and FB is that with FF, the control signal (ie the gain reduction) is calculated by processing the input, whereas with FB, it's calculated by analyzing the output. FB tends to be more program-dependent, which is why sometimes it sounds more musical, but FF can sometimes sound tighter and more controlled.
http://en.wikipedia.org/wiki/File:Compr ... Design.gif
To me, FF1 seems a bit more extreme than FF2 -- generally you get more gain reduction for the same threshold and ratio. I'm assuming it's something to do with how the envelope is calculated, but since they're both feedforward it's hard to say what they're doing differently....
http://en.wikipedia.org/wiki/File:Compr ... Design.gif
To me, FF1 seems a bit more extreme than FF2 -- generally you get more gain reduction for the same threshold and ratio. I'm assuming it's something to do with how the envelope is calculated, but since they're both feedforward it's hard to say what they're doing differently....
- jobbanaught
- mnml mmbr
- Posts: 106
- Joined: Mon Sep 28, 2009 12:10 pm
damagedgoods wrote:The difference between FF and FB is that with FF, the control signal (ie the gain reduction) is calculated by processing the input, whereas with FB, it's calculated by analyzing the output. FB tends to be more program-dependent, which is why sometimes it sounds more musical, but FF can sometimes sound tighter and more controlled.
http://en.wikipedia.org/wiki/File:Compr ... Design.gif
To me, FF1 seems a bit more extreme than FF2 -- generally you get more gain reduction for the same threshold and ratio. I'm assuming it's something to do with how the envelope is calculated, but since they're both feedforward it's hard to say what they're doing differently....
Wanted to know for a long time, thanks for the info
![Idea :idea:](./images/smilies/icon_idea.gif)
ok I need more time then I thought...
but I will talk a bit how I think I could compare them.
I sidechain a sin wave with a kick and look at the resulting wave form (I use a oszilos-mega-scope for this).
you can clearly see the differences with from the different compressor models etc.
example ff2 peak mode:
http://img34.imageshack.us/f/peakff2a00 ... tioinf.png
ff1 peak mod:
http://img85.imageshack.us/f/peakff1a00 ... tioinf.png
the pictures are conform with what I hear too
ps: I need to capture the same screen area in the future, since it's difficult to compare the waveforms otherwise.
but I will talk a bit how I think I could compare them.
I sidechain a sin wave with a kick and look at the resulting wave form (I use a oszilos-mega-scope for this).
you can clearly see the differences with from the different compressor models etc.
example ff2 peak mode:
http://img34.imageshack.us/f/peakff2a00 ... tioinf.png
ff1 peak mod:
http://img85.imageshack.us/f/peakff1a00 ... tioinf.png
the pictures are conform with what I hear too
ps: I need to capture the same screen area in the future, since it's difficult to compare the waveforms otherwise.
-
- mnml mmbr
- Posts: 349
- Joined: Tue Feb 12, 2008 1:38 am
Nice analysis. So I guess FF1 has an exponential attack and FF2 linear (at a guess). Funny though, since I always thought FF1 was more extreme in its gain reduction whereas your plots show the opposite. It'd be easier to compare with a constant sine burst (at 1kHz or so) as your sidechain input rather than a kick drum, and with a longer attack/release time so you can see the shape of the attack envelope. (Depends how much time you've got/how much you actually care.
)
![Wink ;)](./images/smilies/icon_wink.gif)
no. the attack/release curves are defined by the PEAK/RMS/OPTO modes.damagedgoods wrote:So I guess FF1 has an exponential attack and FF2 linear (at a guess).
FF1 just acts way faster than FF2. for example if you set the lookahead of FF2 to 10ms, the results are very similar to FF1 with a 1ms lookahead. IMHO they are exactly the same, just FF1 has more lookahead, which enables it to respond quicker to sharp transients, enabling for a deeper threshold...
ok, nothing scientific proven, but maybe a first finding.
abletons compressor seems to have an exponential release curve and cubase compressor an logarithmic.
logarithmic:
http://img101.imageshack.us/f/cubasesidechain.png/
exponential:
http://img34.imageshack.us/i/peakff2a00 ... ioinf.png/
edit: @damagedgoods, a burst as a trigger is a good idea.
@sk: interesting, I will check that too.
abletons compressor seems to have an exponential release curve and cubase compressor an logarithmic.
logarithmic:
http://img101.imageshack.us/f/cubasesidechain.png/
exponential:
http://img34.imageshack.us/i/peakff2a00 ... ioinf.png/
edit: @damagedgoods, a burst as a trigger is a good idea.
@sk: interesting, I will check that too.
Last edited by regler on Fri Jul 23, 2010 3:56 pm, edited 2 times in total.
-
- mnml mmbr
- Posts: 349
- Joined: Tue Feb 12, 2008 1:38 am
s.k. wrote:no. the attack/release curves are defined by the PEAK/RMS/OPTO modes.damagedgoods wrote:So I guess FF1 has an exponential attack and FF2 linear (at a guess).
Oh yeah, I guess you're probably right on that.
Not quite sure about this. Lookahead generally just means the program signal is delayed relative to the sidechain, so the shape of the envelope is the same but the transient is ducked rather than the bit immediately after it.FF1 just acts way faster than FF2. for example if you set the lookahead of FF2 to 10ms, the results are very similar to FF1 with a 1ms lookahead. IMHO they are exactly the same, just FF1 has more lookahead, which enables it to respond quicker to sharp transients, enabling for a deeper threshold...