Gain staging plugins when mixing

How do others on here gain stage plugins when mixing. Do you compare the peaks or do you use ears, or do you use vu meter etc,

I gain stage my low end to -18 but snares I do -12 and hi hats etc. when I mix I like to get the gain back to -12 peaks with transient material but I don’t know if it’s a good idea since it can be slower way to mix

There is a lot to unpack in this question, so I have been holding off replying hoping Seablade or someone else would get to it.

There are some assumptions in the question that don’t fit my usual inclinations, namely “gain stage” and “plugins” and “mixing” all in the same sentence. I’ll come back to that, but first…

Gain staging is a technical consideration, so the concept of “using your ears” to set gain doesn’t really make sense.
There are really two considerations to take into account when setting gain: the noise floor of the recording channel, and the maximum usable level. With digital recording that is pretty straight forward, because any modern digital device is designed so that distortion stays very low across the entire audio spectrum right up to the maximum usable level, at which point distortion rises very quickly. It wasn’t quite so straight forward with analog tape, or with devices which were transformer coupled, because distortion increased gradually with level, and tended to increase at low frequencies, so you had to make a choice in your maximum input level regarding a signal-to-noise ratio vs. distortion trade off. That trade off is essentially gone with modern digital equipment, you just optimize signal-to-noise ratio with possibly a consideration for safety margin if you are recording live instruments and can’t predict with high accuracy the maximum input level you will have to deal with (the problem of musicians playing more excitedly during recording, resulting in higher maximum level than you got during soundcheck).

That sounds like something learned when dealing with analog VU meters and analog tape machines.
Did you previously work with analog tape, or is that just folklore you picked up along the way but lost the original reasoning behind it?

That goes back to my previous comment about analog tape and transformers having higher distortion at low frequencies. With modern equipment having lower levels for low frequencies has no technical justification, the headroom is identical across the entire spectrum, and if anything distortion is lower at low frequencies than at high frequencies (due to vagaries of how negative feedback works that is tangential to this discussion).

Proper gain staging really starts when recording, well before the mixing stage. Any noise captured at the first stage is a part of the recording from that point on, as well as any distortion from clipping the input, so you want to get that right from the start. Clipping is generally more annoying than slightly more than optimal noise level, so you usually see recommendations to be very conservative with level. Knowing your source helps a lot there, if you know you are recording a single player who is very consistent, you can nudge the levels a little higher (helping signal to noise ratio) than you would want to recording a player who may vary a lot from take to take, or recording a large group where the combined effect of every player being just a little big louder could be a noticeable increase in overall level even though every player only feels like they are playing “a little bit” louder.

For recording, peak level is really all that matters. Knowing the actual peak signal level was a little tricky in the all-analog days, but with digital recording you get that free (with some fine print about inter-sample peaks that isn’t worth pursuing here). So at the recording stage always use peak metering.

I might quibble a little bit about how conservative to be with level setting, i.e. aim for -15 dB vs -12 dB vs -6 dB peak, but this section in the Mixbus manual (and the linked video) is a good overview:
Mixbus manual section on gain staging

Almost all of that will apply to Ardour as well, except for some things which are not always present by default in Ardour, such as the per-channel compressor (described in the section on make-up gain) that is always present in Mixbus, but will only be present in Ardour if you add a compressor plugin to the channel, and the section on tape saturation which is a Mixbus-only feature.

If you have setup your gain staging optimally when recording, there should be very little you need to do when mixing, other than possibly tweaking the channel trim control a little, or moving the region gain up or down a little.

Monitoring peak level is the most appropriate for recording, because what you are concerned about there is not clipping the input, but peak monitoring is not really very relevant to the final mix. It is still somewhat relevant in that you don’t want to clip the output either, which is why a K-meter or similar that can simultaneously show you the peak level along with some type of average level is most useful. The peak display will probably have a hold function, so you can see the highest peak you hit even if you aren’t watching at that instant that the highest peak occurs.

The big insight that Bob Katz had (the “K” in K-meter) was that from a technical standpoint what you need to avoid was clipping on peaks, and because peak level was easy to measure in digital, people tended to just set the peaks at 0 dB FS, or very close to that, but the ear responds to loudness with more of an averaging type behavior, and the ratio of peak level to average level was quite a bit different for different styles of music, and depended a lot on how heavy handed the compression and limiting were used on a particular recording. Because of how those two facts interact, you ended up with a wide variation in perceived loudness between different recordings with very similar peak levels, and you also ended up with a downward spiral in the quality and dynamics of recordings as recordings tended to have more and more compressed dynamic range in an attempt to have increased perceived loudness within the technical limit imposed by a fixed peak digital level.

The solution proposed by Mr. Katz was to use the same technique that the film industry imposed years ago when film soundtracks got into a loudness race, which was to use a fixed digital level to monitor SPL gain, and then mix so that levels sounded comfortable at that calibrated monitor setting.
You can go read the original papers at DigitalDomain articles if you like, or linked from the Ardour manual section on metering, and I would highly recommend that, but the short of it is watch your peak levels when recording, because avoiding clipping distortion is important there, but when mixing watch the average levels on the output bus, and mix by ear in a way that keeps the average levels in the green, which you will probably do naturally if you calibrate your monitors properly.

Thanks for your reply, I will look more into this

You should also re-read the answers you got in May last year, when you asked the same question.

Seablade these days is incredibly busy (2 jobs and running a business simultaneously does that to you) and often doesn’t have a whole lot of time dedicated to answering questions anymore sadly so misses a lot he would like to.

He also seems to have taken up speaking in the third person for some reason.

That being said, @ccaudle 's response is a great detailed response to the topic that probably covers a lot more than the original question was meant to ask (And that is a good thing:)

There are very specific (Much more the exception than the rule) exceptions to this. Typically it comes when a plugin is designed to operate in the way desired on signals that are not similar in level to the ‘average’ for whatever reason. So my real quibble would be that I would add to the following:

a third criteria, the desired operational level for lack of a better descriptor, which is essentially the level that gives you the sound you want out of the plugin. For many plugins this ‘ideal nominal level’ will be anywhere between -25dBFS and -12 dBFS (-18dBFS in particular in my experience is often a good place to sit for most modern digital processes in the 24 bit realm) so the first two considerations @ccaudle mentions would be all you have to worry about, but in very specific cases (Specific plugins, or non-standard use cases of other plugins) it may not and you would need to stage to this effect. This is most common when you are looking at add a form of noise or distortion to the signal. This is also where ‘staging by ear’ is something that can and should happen, but again I will stress this is very much the exception rather than the rule, and otherwise I would stick to @ccaudle 's advice which is intended to keep as clean a signal as possible throughout the chain, and correct in the vast majority of cases.

     Seablade

Thanks die responses,I may have forgotten I asked this question before so I apologize for that, however I did want more input as I understand many people mix different and get results there looking for. No I t has been a serious challenge but I’ve managed to resolve lots of the problems I’ve been having in regards to mixing, and signal chain etc.

I do like hearing about others workflows as it helps me to get different perspective on how I do things. M
Thanks!!