Ozone 8 maximiser clipping in Logic x

iLogical

a member member
So I'm trying to bring my track to up to about 14LUFS (more out of interest than anything since this is on the output of the project rather than from an already bounced file in a separate mastering project).

I have recently picked up this Ozone plugin with a maximiser which is very nice looking and seems to sound fine so far but while playing around with it I'm finding some odd things.
My track is Peaking around -2.4db without anything on the output.

However, if I set the Ozone up to bring me to 14LUFS it clips at about 1.6 on my output in the logic mixer.

I have the "Ceiling" set to -1 to try to avoid any digital clipping etc but it makes no difference. I still clip up above 1db on the logic stereo output.
I tried even using the Ozone mastering AI to see if it would figure out something clever since I'm not especially familiar with this yet but no luck - same problem.
There is absolutely nothing else on the output (not even a meter after the ozone).

I was under the impression that the maximiser was meant to act like a limiter so setting the ceiling should kill the sound and make it horrible and squashed before it would allow me to clip on the output...
Why is this not the case??

Any advice about where I am going wrong would be much appreciated.
The image is of the settings chosen by the AI

Screenshot 2019-08-12 at 13.09.11.jpg
 
Click on the Learn Threshold “switch” and maybe even the True Peak one and just let your mix play the entire length of the cycle region.

P.S. (edit) if your track is already peaking at -2.4dB before using the Maximizer, it's probably a bit too hot already.

The Maximizer AI is only going to work if you pick the right section of your mix for it to sample, so it's going to depend entirely on whether that section is representative of the entire track. Unless you're making sausages, it's unlikely to get you to -14dB LUFS on the nose, or even within a dB. But, with practice you'll get close enough and then you just need to tweak the Threshold to get it closer, though if your first pass is off by several dB, I'd suggest resampling with the AI.
 
Last edited:
Hey Keith,
Thanks for your reply - good stuff to think about although I must confess it doesn't solve it.

This is the loudest spot in the track and yes I did the full length of the track thing with learn threshold already as part of my testing the plugin - that was before I focused on this particular problem section.

A couple of questions though...
I'm interested that you say -2.4 is too hot? I'm no mastering engineer but I've been mixing for years and always received good feedback from the mastering studios I use. I have always heard that somewhere around -4 to -2 is ok for a mix which is off to be mastered. Where would you put your track? BTW I'm not saying it's -2.4 on average... -2.4 is the highest peak in the track.

Also correct me if I'm wrong but regardless of where I set the threshold this Ozone maximiser should behave as a limiter and stop at (or just above allowing for digital peaks) the ceiling which I set?
This is how my other limiters work - For example: I can chuck a hell of a lot at my L2 plugin before it goes above -0.9 if I set the ceiling to -1.0. Sure it sounds horrible but it is not a compressor which should let transients pass to the ratio you set, it is a limiter and the ceiling should be (basically) an absolute ceiling yes?

Maybe I have misunderstood what a "Maximiser" is????

Edit: I am really not fussed about a perfect 14LUFS just using it a rough guide to see how Ozone 8 sounds when I push it to a fairly standard loudness.
 
Perhaps the Logic meter is accounting for potential intersample peaks using a different prediction algorithm than the Ozone processor.
 
Hey Keith,
This is the loudest spot in the track and yes I did the full length of the track thing with learn threshold already as part of my testing the plugin - that was before I focused on this particular problem section.
If you don't set "true peak" the maximizer can (and almost always will) create an average loudness that will result in intersample peaks above the limit ("Ceiling") setting. Especially if you are bouncing to MP3. The worst case I've seen (bouncing to MP3) has allowed maybe a handful of peaks over 0, but typically I get a maximum reconstructed peak below -0.5dB when measured with something like Orban.

If you just run the AI over the "problem section," what are the resulting Threshold and Ceiling values (with and without True Peak) set? (I suppose it's possible you have a mix that cannot be raised to that -14dB LUFS strictly via the Maximizer Assistant, e.g., if the dynamic range is too high. In the full product, the Assistant will typically introduce a both a Compressor and Dynamic EQ component in front of the Maximizer that should compensate for that.)

A couple of questions though...
I'm interested that you say -2.4 is too hot? I'm no mastering engineer but I've been mixing for years and always received good feedback from the mastering studios I use. I have always heard that somewhere around -4 to -2 is ok for a mix which is off to be mastered. Where would you put your track? BTW I'm not saying it's -2.4 on average... -2.4 is the highest peak in the track.
That's just me - I have pretty simple mixes. If it's been good, then please ignore my thoughts on that topic!

Also correct me if I'm wrong but regardless of where I set the threshold this Ozone maximiser should behave as a limiter and stop at (or just above allowing for digital peaks) the ceiling which I set?
This is how my other limiters work - For example: I can chuck a hell of a lot at my L2 plugin before it goes above -0.9 if I set the ceiling to -1.0. Sure it sounds horrible but it is not a compressor which should let transients pass to the ratio you set, it is a limiter and the ceiling should be (basically) an absolute ceiling yes?
Only if True Peak is turned on and you bounce to a non-lossy file IME.

Maybe I have misunderstood what a "Maximiser" is????

Edit: I am really not fussed about a perfect 14LUFS just using it a rough guide to see how Ozone 8 sounds when I push it to a fairly standard loudness.
Well, it works for me, is all I can say, though I am not using Elements, but the full Ozone 8 product, which includes other components.

P.S. (Edit2) I see that you are running this on the mix. I'd recommend bouncing to just a non-lossy stereo (un-normalized, un-dithered, 24-bit), and then create a separate project for mastering (or do this in an Alternative). I'd be curious to see what the integrated LUFS and true peaks (using Logic's utility plugins) of the un-mastered file are before anything is done with Ozone.
 
Last edited:
Thanks again for more thoughts!

If you just run the AI over the "problem section," what are the resulting Threshold and Ceiling values (with and without True Peak) set? (I suppose it's possible you have a mix that cannot be raised to that -14dB LUFS strictly via the Maximizer Assistant, e.g., if the dynamic range is too high. In the full product, the Assistant will typically introduce a both a Compressor and Dynamic EQ component in front of the Maximizer that should compensate for that.)

The image I posted is with it having just used the AI on the problem area (hence the cycle) and yes... it automatically dipped the ceiling to -1 because I reset it and tried again after to check it handn't just kept my manual ceiling.

-------------------------------------------
So I think I have a rather anti-climactic resolution which is that somehow the logic output meter was just plain wrong. The reason for this conclusion is that I decided as an experiment to just chuck in some of my usual rough master plugins which consists of an eq 2 compressors for staging and the L2 as a limiter (plus a LUFS meter after everything). nice and simple. I bypassed the Ozone and brought the LUFS up to 14 using the compressors as expected the L2 was letting through peaking at -.9 when set at -1 (due to inter-sample peaks) and that was that. Then I switched back to the ozone and disabled all those other plugins......and I got the same result: -.9 with the ceiling -1.

So my very disappointing conclusion has to be that is was just glitch!!!! AHHHRG! haha

Still good to get the old HRF troubleshooting in on it.

Occasionally the bug re-emerges and seems to be solved by adding (and removing if you want) another plugin of any kind on the stereo out. Rather odd but predictable at least!

The final piece of proof is that even when the Logic stereo out says, for example, clipping at 1.6 the logic multimeter is still saying it is only at -0.9. So yeah a silly little metering glitch.

Cheers for the help!
 
Good conversation .. I've found the Elements ai to be pretty helpful for eq suggestions on a track that I've already mastered , but I always tweek it. I've found the hardest thing is finding the loudest part of a song for the assistant to use. I've found the best way is using the "learn" while reading short term lufs, when the learn settles down during the loudest part , stop transport and turn Off learn. It only listens for a few seconds before it makes its settings. The documentation help section ,helped a little... mark
 
Good conversation .. I've found the Elements ai to be pretty helpful for eq suggestions on a track that I've already mastered , but I always tweek it. I've found the hardest thing is finding the loudest part of a song for the assistant to use. ....
I've never used Elements, but on the full version, you can just turn on Learn and let the entire song play through, i.e., if you're trying to get a Maximizer setting for the whole piece that hits your target LUFS. Then, measure the result with Orban or whatever you've got, though I generally do the measure after getting close and then simply adjust the threshold to get it within a couple dB and call it good. If I see that hitting the target significantly changes the dynamics or waveform (a lot just shaved off), then I know i have to go back to the mix, or at the least do some tweaking on the mixbus compression.
 
The limiter is not designed to never clip really. Here's a vid that give's a better explanation than i can:
YouTube

I can say that the limit you set is dependent on the delivery medium. For instance , the -.3 db is supposed to be "CD" level. Also the IRC type is very critical in getting a good finished result.
 
Back
Top