Maximizer v. Limiter (v. Brickwall Limiter)?

What’s the difference between Maximizer, Limiter, and Brickwall Limiter? They all do the same the same thing in my mind, so I must be missing something, since Cubase has three separate Dynamics Plug-ins for them.

The Cubase 8 Plug-in Reference manual says,

“Brickwall Limiter ensures that the output level never exceeds a set limit.” (p. 38)

“Limiter: This plug-in is designed to ensure that the output level never exceeds a set output level, to avoid clipping in following devices.” (p. 47)

Those descriptions seem virtually identical to me, so how should I decide whether to use the Limiter or the Brickwall Limiter?

And then there’s:

“Maximizer: This plug-in raises the loudness of audio material without the risk of clipping.” (p. 48)

Isn’t that exactly what the limiters do? I’ve always thought of limiters as allowing me to raise the gain (which raises the loudness) while ensuring that peaks don’t clip.

Why does Cubase have three plug-ins for this, when they all seem to do the same thing? What am I overlooking?

In the broadest terms a limiter restricts the peak levels in audio material whilst leaving lower level material untouched, whereas a maximiser raises the average level of the material (high and low level) and restricts the peak levels.

My understanding is that in general limiters behave like compressors with high ratios and high threshold, pulling back peak levels. Brickwall refers to that ratio being infinite.

For example, a compressor set to 20/1, for 20db input gain above threshold knee, there is 1db output gain. A steep slope.
A compressor with ratio set to infinity/1, for 20db of input gain above threshold knee there is 0db output gain. Would look like a vertical line on a graph, hence brickwall.

Hope this helps, no doubt someone else will be able to put it better than me, but that’s the principle.

This makes sense to me.

This makes sense to me too.

The thing is, if you use a limiter to squash the top 10db, and then you raise the output gain by 10db, this will raise raise the average level (high and low level) while restricting the peaks; so I still don’t understand how that’s different from a maximizer:

I appreciate your explanation, but I’m still confused.

If you compare the limiter and maximiser plug ins by looking at the output audio waveforms in comparison with the input waveform. The limiter will leave the waveform untouched until a peak occurs in which case it will dip the signal level. A maximiser will raise the entire level of the waveform and then apply limiting to the peaks.

So you could think of a maximiser as two components an amplifier followed by a limiter. Or, I guess vice versa.

Thanks again. No need to add an amplifer after the limiter, as the limiter includes an “Output” dial for adjusting the output volume. It seems strange to me that Steinberg added a whole different plug-in and named it “Maximizer”, instead of just telling people, “increase the Output level on the Limiter if you want to raise the volume”.

Maximisers are more complex that the multi-linear segments of compressors or brickwall limiters, though they are effectively providing both those functions.

I haven’t used the Cubase Maximizer, but the Ozone one allows me to specify the range of levels over which it is to work. It has several algorithms, with the most intensive bringing my old system to its knees.

However, it is still better to optimise levels by judicious mixing and sculpturing with automation, than aggressively use a maximiser, as the latter can get quite harsh when pushed. They are good to just give a gentle few dB boost of overall levels.

Thanks. Steinberg chose to constrain their Maximizer documentation to five somewhat cryptic lines, so I’m going to stick with my less mysterious dynamics processors.

Despite the ‘economy’ of the documentation, a well applied maximiser can give results that plain dynamic processors would have difficulty duplicating. Try some of the presets to see what is offers.

Just remember, any maximiser or brickwall limiter MUST be placed post-fader and be the LAST plugin before rendering via export.

Unless you are exporting a lower bit depth than what your project was mixed at!
If you mixed at 32 bit float or 24bit and are exporting down to 44.1 kHz 16 bit or less than 192kbit MP3 then you would want the UV22HR Dither Plugin to be in the last Post-Fader slot.

Sorry, I forgot that Ozone’s Maximizer has built in dithering, covering both in one plugin.

Yes, if dithering is a separate plugin, then the order is:

  1. Maximizer
  2. Dithering.
    Both post-fader of course.

Perhaps another way of appreciating the difference: A Maximiser placed on an acoustic guitar really thickens it up nicely. And by automating the Optimise you can add in just that little bit of occasional punch. I wouldn’t use a limiter for that.

Whether or not I ought to be doing that post-fader is something that Patanjali has just reminded me (thanks)…

Also, there is some talk here of squeezing dbs so you might find this thread relevant: http://www.steinberg.net/forums/viewtopic.php?f=198&t=73412

Thanks for the link.

In light if this thread, I’m thinking of it as a soft-knee compressor, with substantial distortion, and makeup gain. Automating it sound interesting; I hadn’t thought to try that. Thanks all.

A Maximizer is just a brickwall limiter that automatically pulls up the gain. That’s all. It just makes life easier but is essentially the same thing.

Alistair

It depends on what you are doing. I use maximizers (or brickwall limiters depending on how you want to call them) all over my mixes on individual channels. I want them to be pre-fader so that the fader adjusts the level of the entire channel, not the level going into the maximizer.

Actually, I would love an option to turn all the the post-fader slots into pre-fader slots. If anything, the whole post-fader slot system is a nuisance to me. Even nicer would be if it could be done on a channel by channel basis.

Oh, and more slots please. :slight_smile:

Alistair

I did think of that afterwards, as they can of course be used anywhere, but since they can be quite CPU intensive (maximisers, that is, not brickwalls), using them in too many places can crush a system.

I have never come across a situation (that I remember) where I used a post-fader slot on a non-output track. Would be interested to hear of a use case.

I would suggest that all slots for all tracks, except the outputs, be pre-fader, but with the option, on a per-track basis, to progressively switch to post-fader, last one first.

The default for the output tracks should remain post-fader on the last two, as I suspect that many would too easily forget to switch to post-fader when they insert maximisers and ditherers, since they MUST be post-fader.

I am used to using Ozone’s Maximizer, which incorporates dithering, so ensuring no clipping. It even has inter-sample clipping prevention.

However, I just thought of a possible problem when using a separate maximiser (or brickwall limiter) and dithering plugins.

Of course, the maximiser is placed first, to stabilise the signal range for the ditherer.

But if the maximiser sets the peaks to zero full scale (0FS) or just below it, wouldn’t the noise that the ditherer adds to do its thing be likely to exceed 0FS, so undoing the effectiveness of the maximiser?

Do ditherers include the micro-dynamics adjustments to prevent this, or do their basic algorithms, by their nature, prevent it?

The latter is closer - “their basic algorithms, by their nature, prevent it”
You can still have small overs using a dither plugin.
This is why most wouldn’t peak limit at 0dbFS but at -0.2dbFS instead to allow that slight headroom - especially useful when converting to MP3.
In any case, dither is ALWAYS the final ‘MASTERING’ operation on a signal path before the final format ‘print’.
In otherwords, you shouldn’t dither a ‘mix’ unless you are printing your reduced bit depth MASTER straight out of the mix… as sometimes happens with EDM.

I was under the impression dither only affects the least significant bits, never altering the more-significant bits. It doesn’t just add any random noise (which could cause clips); it rather adds specialized noise devised with respect to your signal so that it changes only the least significant bits.

I could be wrong, though.

Sounds like some smart people involved there. So it would appear to come under the ‘by its nature’ category, rather than micro-dynamics.

Except, I’m googling for info to substantiate my suggestion, and I’m not finding anything, so I may well have been totally wrong.