Question on mixdown process - turning the levels down

[nowaysj]

And some plugs are 24bit and will clip, and what sunk said, and axe, you left out a very important 'not'.

[AxeD]

When I mix in a 32 bit float system, I don't care as much whether my Oxford limiter's input is clipping.
That's what I meant I usually do proper gain structuring also within a daw, but in theory there's
quite a bit of headroom in floating point.

But yeah, it completely depends on the type of plugins and what you're trying to accomplish.

[hubb]

With all this fine advice being handed out, you still have to go and have a look on the bounced out or fully rendered waveform. See if any curves are being squared off, because digital metering can never actually be trusted. We all want a digital meter to be a physical readout, but in nature it's just a very simplistic graphical representation.

[rareclub]

^ you could just use smexoscope, or some other oscilloscope of your choice.

http://bram.smartelectronix.com/plugins.php?id=4

[hubb]

No, that's also a digital meter. Same issue, despite it doing a live interpretation of a waveform.

[SunkLo]

A rendered waveform is the exact same thing. There's no difference between an algorithm that draws a waveform from blocks in real time, and an algorithm that draws a waveform based on a file. There's meters that detect inter-sample peaks, which is primarily what you need to look out for. Just leave some extra headroom so you don't clip your convertors or mp3 compression algo. Not everything needs to be normalized to 0.0dB

[hubb]

The problem with smexoscope is you can't zoom in enough to judge the squaring off properly. And the fact that it is real time, means there's a form of interpretation /compensation going on, either in the form of the info it gives or the info it recieves.
Im not that technical, but maybe the process power needed in real time affects the accuracy of that algorythm?. Just spit balling here, there's no doubt it's not a 1:1 physical representation of sound. There's dynamic drifting, there's peaks - there's psycho acoustics etc.

But there was a genuinely good point to what I was saying that got lost in all this correctness. Something forums has a problem with, so let's get back to that.

Use the ears and get a blindfold. Maybe the same kind as tmnt has, because then you can peek out of it when needed.

Oh and btw, I enjoyed a mix of yours (?) that happened to pop on my sc. Nice one.

[nowaysj]

You're a dirty boy though hubb. I'm trying to be good. I suppose we have to strive towards the opposite of our nature? I tend to get myself into trouble, so following all these guidelines and what not, saves me as much trouble as it can once shit gets organic.

I'm not comfortable clipping things earlier in the process. Yes you can do it transparently, at that time. But I've had issues where further down the line when those sounds are mixed with other sounds and then later finally processed, like master processing, that transparent becomes less so, and you're like, oh that is that fucking sound I clipped a week ago.

I suppose I could get a couple more db by clipping, I'm at that point to try, do you recommend a clipper, or do you just take it up to digital max and then take it down again?

[SunkLo]

Real time waveform display isn't that intensive at all. The plugin gets an array of double values and just plots them on screen. It's a perfect representation of the sampled audio. It's not a completely perfect representation of what's coming out of your speakers, but neither is the rendered waveform, they're both discrete. Sometimes some ultra fast peaks can sneak through in between samples, which is why you need something that detects inter-sample peaks. That's just a meter that upsamples to raise the upper frequency limit and catch faster movements. It's usually not that big of a problem if you leave a few dB of headroom.

Hmm a mix of mine you say? If it was under 'SunkLo', that's Boddika and Joy Orbison's record label that jacked my name.

NWJ: Clipping shit is way easier with a clipping plugin than trying to do it in a DAW. You have to render out audio for it to actually clip in a 32 bit engine, which makes it a pain in the ass to get the clipping set right. It's most transparent if it's just the transients being clipped. That scenario we were talking about earlier, where the transients getting through a heavy compressor are super loud, is a perfect situation to use clipping. Just squish it up nice n good and then your transients will be much easier to hit with the clipper because of the higher level. The compression should keep the body from riding up into the clipper's hot zone. You might wanna have some soft clipping before it completely squares off. Waveshapers are pretty good at dialing this stuff in.

[hubb]

Yeah.

I feel like anything considered destructive editing, like something rendered that clipped, well you're forced to either accept it or let it go. So if you're at that spot where the composition and design is done and you have to spice it up, anything destructive is sort of freeing in the sence that it will either work or not, so you just try it out.

I have this whole silly idea of what anything with a threshold presents when its running on top of your 'rollercoaster', but It's probably better if I want to be taken 'seriously' that I just say, it's another variable that easily numbs our perception. Like women levels at that.

[nowaysj]

already snorkeled it already - I cannot be held accountable for my language. In fact, I should be applauded for the extent to which I can communicate right now. I've crashed myself upon the rocks of another song that is just getting away from me. I'm pissed as fuck, delirious, and dim whited, and now need to go try and work on it some more.

[hubb]

Real time waveform display isn't that intensive at all. The plugin gets an array of double values and just plots them on screen. It's a perfect representation of the sampled audio. It's not a completely perfect representation of what's coming out of your speakers, but neither is the rendered waveform, they're both discrete. Sometimes some ultra fast peaks can sneak through in between samples, which is why you need something that detects inter-sample peaks. That's just a meter that upsamples to raise the upper frequency limit and catch faster movements. It's usually not that big of a problem if you leave a few dB of headroom.

I know you are correct. My feeling is that our perception of loudness is so much richer in expression and even in volume between just two not too different sounds than what a digital meter read out will us make believe. The reason is mostly the difference in frequency distribution within those two sounds and our ears frequency response to them. And here's my point, we as producers tend to then change one of those two sounds (or put a limiter on the master), because it makes sence numerically. And in sequenced music, numerical can be seen as control or even expression, but it kills the experience of the individual sound and the richness that I believe is actually louder. Is it ?

Hmm a mix of mine you say? If it was under 'SunkLo', that's Boddika and Joy Orbison's record label that jacked my name.

Ahh ok, thought it was a djs name. was definitely that kind of vibe.

[hubb]

already snorkeled it already - I cannot be held accountable for my language. In fact, I should be applauded for the extent to which I can communicate right now. I've crashed myself upon the rocks of another song that is just getting away from me. I'm pissed as fuck, delirious, and dim whited, and now need to go try and work on it some more.

No, no I wouldn't put it there if it wasn't resounding profoundly within this block of meat above my shoulders.

[bennyfroobs]

Many MANY plugins will give you 2-4db gain to their presets. I think this is very dishonest and is an attempt to deceive you. Fuck that. Unless I'm specifically using a plugin to increase my gain, I want signal peak in to equal signal peak out.

yeah i h8 that shit its so fucking annoying!!

[bennyfroobs]

okay, read the whole thread now

bigup sunk & mr jose, i always learn some intersting stuff when u guys are gabbing about proudction biz

i ahve questioms though

1 what is 'latency'? i know it as "ping" from videogames which is like the time it takes for ur pc to send info to the masterserver and back. is it similar to that?
2 what does 'transparent mean?
3 what is 'clipping' and 'soft clipping'?

[bennyfroobs]

4 did they really steal ur name. what ploppers

[SunkLo]

1 what is 'latency'?

Yeah just when a plugin introduces delay. Like a look ahead limiter delays the audio output by a few samples so it can react instantly to peaks.

2 what does 'transparent mean?

Not obvious sounding.

3 what is 'clipping' and 'soft clipping'?

Clipping is hitting the digital ceiling where your audio can't go any louder. The waveform gets squared off on the top and you get nasty audible distortion. Soft clipping is similar but instead of immediately hitting the ceiling, your audio gets gradually more squished as it approaches the clipping threshold. Since soft clipping isn't as abrupt as hard clipping, it doesn't generate as much distortion.

4 did they really steal ur name. what ploppers

Yeah they even have the exact same capitalization. I sent a tweet to Boddika on twitter saying "hey you lot stole my name" with a link to my dsf profile and join date, but it got ignored.

[nowaysj]

Latency - some plugins require a small delay to do their processing. It can be anywhere from a sample to 2000ms. I think I remember some psp stuff from back in they day with redic latency. So if use a processing plugin that introduces latency, that channel is going to be late by however latent the plugin is. That channel will fall out of time. Or if you are working in parallel - splitting say a channel into two and processing one of those two, it will fall out of time with itself and phase.

Daw's got their shit together and implemented APDC - Automatic Plugin Delay Compensation. It works differently in different daws, but essentially the daw/host receives a message from the plug about how slow it is, and delays all other tracks by that amount. Sounds simple, right, but when you start bussing etc, it can get hells of complex.

Fl's apdc is not flawless, as apdc seems to never be flawless. There are plugins that don't report, or not in the right way. Whatever. Usually it works, sometimes it doesn't.

I skip the issue by using plugs without latency, but if I do use latent plugs I manage it my self. Which is a limitation.

Transparent means just that. When you look at your wall it is not transparent. When you look at your window, it is transparent.

But as maybe you are guessing, your window is not perfectly transparent. The light passing through the window is distorted, to some extent.

Same for plugs. A truly transparent plug has no effect on the signal. Signal in = signal out. They should null if the polarity is reversed.

So when discussing transparency, we're talking about how much distortion a processor introduces into the signal. How much signal in differs from signal out.

When speaking in terms of limiting or compression. We're talking about how much the dynamics can be changed while still SOUNDING like signal in = signal out of the device. FabFilter's Pro-L is relatively transparent, imo. It can chop peaks off signals fairly aggressively without introducing a lot of distortion. Or the distortion it does introduce is pleasant.

When you look at a waveform, like a snare. There are waves going up and down. They are round at their peak. A clipper chops the peak off the signal/wave and leaves it flat. If you do this in the right way, to the right extent, that peak can be removed without significantly altering the sound.

And google is your friend. I also expect SunkHigh will be in shortly to correct my sloppy non-technical use of all these terms. So pay attention.

[nowaysj]

Nope.

[SunkLo]

Heh