Downward vs Upward Compression

VST, AU, AAX, CLAP, etc. Plugin Virtual Effects Discussion

What type of compression do you prefer ?

Downward Compression
30
83%
Upward Compression
6
17%
 
Total votes: 36

RELATED
PRODUCTS

Post

Dionysos wrote:upward compression will also always affect and shape peaks/transients.

How can that be ?

Post

Upward compression can be great but it can easily become a nightmare when it comes to guitar feedback or system noise. I like to use it to bring out the sustain and body of things such as plucked strings or electric pianos. Also can bring forward the attacks/decays of pads.
futurefields wrote:Does NY compression qualify as downward or upward compression?
Certain particular conditions can technically bare identical results from the two, but ultimately they are different. In order for the two to be the same, there would have to be no non-linearities in the compression algorithm and envelope times would have to be slow enough to not change amplitude within a single cycle of the present frequencies.

It might not seem intuitive until you compare the similarly effected signals side by side.

In my case I use heavy parallel processing but rarely by one effect at a time, I.E., I won't just mix the wet and dry of a compressor but I will mix one chain of compression, distortion, and EQ with another of the same effects with very different settings, so the results would never really resemble upward compression.

Post

VibraSound wrote:
Dionysos wrote:upward compression will also always affect and shape peaks/transients.

How can that be ?
Just as the release time of downward compression will affect the 'sustain' signal, the release of upward compression will affect the peaks.

You seem to have Maximus, so you can use that to try it out for yourself. Upward-compress something with it and watch how the gain envelope doesn't immediately go back to 0dB when a peak hits.
Also, not all peaks will start with a full-on signal, but instead start quiet and take some time to hit the highest peak. It's quite hard to not mess up this kind of peaks with upward compression, but of course it can be used for creative purposes.
The Chase wrote:I will mix one chain of compression, distortion, and EQ with another of the same effects with very different settings
Doesn't that create quite obvious phasing?

Post

I don't know what New York Compression is, but i'm hearing what seems to me to be alot of non-sense speak on upward compression.
Compressing the signals upwards is just taking all the fading signal amplitudes falling below a determined level, and holding them there or bringing them back up to that point for a predetermined, but limited length of time.

Signals fading away are moving slowly compared to attacking signals.
Words like peaks,release, just don't apply like they do with downward comp.

With upward comp you get the effect of increasing the loudness, since your not letting go of signals that would have disappeared naturally. So the average signal level increases.
The peak signal level is un-effected, as was stated earlier.

Sure, it's inversely similar to a downward compression's mechanics, but it doesn't need to handle signals on the attack, but the ones on the decay, and they are changing much more slowly over time...so the processing is minimally taxing. As a result, way less detrimental to the signal quality as well.
But again, it's apples and oranges.

Which type of compression method you would choose depends on where you have an exaggeration, or imbalance of signal dynamics between the peak and average signal content. Thus using compression as a correctional process.
Processing to effect the signal in a way as to enhance is another thing entirely, and a matter of artistic creation. 8)
....................Don`t blame me for 'The Roots', I just live here. :x
Image

Post

annode wrote:The peak signal level is un-effected, as was stated earlier.
Would be nice if it worked like that in reality, but I've yet to hear this being true.
If there's upward compression happening, and a loud peak arrives, this peak will also be amplified above its input level for a short time before the gain is released back to 0dB. You can only avoid this by using a very short release time, but we all know the downsides of short releases when compressing.

Post

Dionysos wrote:
The Chase wrote:I will mix one chain of compression, distortion, and EQ with another of the same effects with very different settings
Doesn't that create quite obvious phasing?
Nope, not unless there are timing problems :)

Post

Downward compression makes weak sounds louder as well, especially when compressing hard.

Post

Dionysos wrote: But it isn't. If you downward compress a signal and then add that to the original signal, you're still adding downward compression to the signal, nothing else. ;)

If there's a difference between upward and downward compression, it's how the gain envelope is created, and this doesn't change only because you mix dry and wet.
It amazes me how many otherwise very intelligent people cannot grasp this concept. There is some strange way to overthink the problem.

Post

I can see where this is all coming from ,I think.
In the book, "Mastering Audio - the art and science - by Bob Katz , he talks about upward compression. His description of the process mechanics is suggested to be "parallel compression". In a "parallel arrangement", he has a downward compressor on one side, and a sample-acurate delay-line on the other side, with both outputting to a single buss.

Those two signals are mixed, the compressed signal and the dry, delay compensated signal. (The compressor's look ahead will create latency to some degree.)
To get the up-compressed outcome, the compressor is set with it's threshold down around -50dbfs. Ratio is set about 2:1 - 2:1.5. Release between 250 - 350ms. Or as much a 500ms for lingering reverb tails, in order not to emphasize them too much by bringing them back up into the mid level range. (this is not verbatim.) Attack is set as fast as possible. (1ms or less)

The down compressor reduces the level of the high to mid level program audio, leaving the lower to lower mid level virtually un-effected. (after the gain control is increased enough for a mix of wet and dry to the desired dynamic effect.)
When this audio is mixed with the dry, un-compressed audio, the result can be, increased levels from what was the range between low to mid to as much as 6dbfs. Because the audio in that range is being doubled.
The compressor is reducing or almost eliminating the high to mid levels within it's signal. When the two sides are mixed together the high energy, peak signals are untouched since they are derived from the dry side only.

Whether the guts of any upward compression technique works this way internally, I don't know. It appears that some here are thinking this way.

PS: I have to say...it appears to me the downward compressor is actually a limiter.
....................Don`t blame me for 'The Roots', I just live here. :x
Image

Post

The Chase wrote:
Dionysos wrote:
The Chase wrote:I will mix one chain of compression, distortion, and EQ with another of the same effects with very different settings
Doesn't that create quite obvious phasing?
Nope, not unless there are timing problems :)
Just watch out with mixing an EQed signal with the same, but differently filtered or dry signal. With standard EQs there will normally be phasing.

Post

The Chase wrote:
Dionysos wrote:
The Chase wrote:I will mix one chain of compression, distortion, and EQ with another of the same effects with very different settings
Doesn't that create quite obvious phasing?
Nope, not unless there are timing problems :)
gol wrote:you should never send to an IIR filter, only insert

Source : http://forum.image-line.com

Post

Dionysos wrote:
The Chase wrote:
Dionysos wrote:
The Chase wrote:I will mix one chain of compression, distortion, and EQ with another of the same effects with very different settings
Doesn't that create quite obvious phasing?
Nope, not unless there are timing problems :)
Just watch out with mixing an EQed signal with the same, but differently filtered or dry signal. With standard EQs there will normally be phasing.
Yeah, it's kind of like the Swamps of Sadness in The Never-Ending Story. If you don't believe in it, it can't really hurt you. Image

Practically though, it can be very useful for shaping drums. It is very popular in DnB/Dubstep and other types of electronic music. For instance if you send drums to two tracks, limiting one signal on one and use a compressor to bring out the attacks on another, EQing becomes a lot more dynamic - EQ-ing the the first track more effects the body/resonance of the drums while EQing the other effects the transients, allowing you (for instance) to add more low-mids to the attack to make them more punchy without also boosting this area in the body/sustain of the drums which can make them too boomy/boxxy. Same deal with distortion - you can overdrive the crap out of drums, mixing the wet and dry, but also boosting/cutting exactly what you want/don't want from each.

Post

Nystul wrote:It amazes me how many otherwise very intelligent people cannot grasp this concept. There is some strange way to overthink the problem.
I can live with this.

Simply reversing a process doesn't make it simple.

So says I. :)
We escape the trap of our own subjectivity by
perceiving neither black nor white but shades of grey

Post

annode wrote:PS: I have to say...it appears to me the downward compressor is actually a limiter.
How so. Isn't a limiter actually working on peaks, not ignoring them?
We escape the trap of our own subjectivity by
perceiving neither black nor white but shades of grey

Post

VibraSound wrote:
The Chase wrote:
Dionysos wrote:
The Chase wrote:I will mix one chain of compression, distortion, and EQ with another of the same effects with very different settings
Doesn't that create quite obvious phasing?
Nope, not unless there are timing problems :)
gol wrote:you should never send to an IIR filter, only insert

Source : http://forum.image-line.com
He said quite obvious phasing. Of course theoretical phasing will always be there with an EQ. But as far as the end result is concerned it's not really an issue. I'm sorry that I did not tack down that semantic issue. I invite everyone to try it and see.

Post Reply

Return to “Effects”