ReFx Nexus 3 (N3) Finally here!

VST, AU, AAX, CLAP, etc. Plugin Virtual Instruments Discussion
Locked New Topic
RELATED
PRODUCTS
Nexus

Post

quantum7 wrote: Thu Nov 14, 2019 11:50 pm I'm a little late to this party, but for some reason I thought that Avenger was Nexus 3. I tried out Avenger and a lot of the expansions last month at a friend's studio and was pretty impressed. I had Nexus a few years ago, but got rid of it. IMO Avenger was soooo much nicer than Nexus 2. What do you think that Nexus 3 will have that Avenger doesn't already have? They seem like they both cater to the same buyers.
I'm guessing it will retain the Nexus interactivity but with some of the Avenger specs. Meaning, it will be more about the sounds in relation to making songs and performance and less about spending hours editing or creating patches.

Post

steve.lindqvist wrote: Fri Nov 15, 2019 12:01 am I am a pro indeed:

Nexus user = Pro
No Nexus user = No pro
:hihi:
:wink: :tu:
Image
Intel® Core™ i9-9900K•Cubase 11•Presonus Eris E8 XT•Focusrite Scarlett 18i20 & Octopre•NI Kontrol S61 MK2•Stein­berg CC121•Synthesizers: Arturia Casio Korg Roland Yamaha

Post

quantum7 wrote: Thu Nov 14, 2019 11:50 pm I'm a little late to this party, but for some reason I thought that Avenger was Nexus 3. I tried out Avenger and a lot of the expansions last month at a friend's studio and was pretty impressed. I had Nexus a few years ago, but got rid of it. IMO Avenger was soooo much nicer than Nexus 2. What do you think that Nexus 3 will have that Avenger doesn't already have? They seem like they both cater to the same buyers.
I think this quote from a parallel topic is quite compelling:
noiseboyuk wrote: Thu Nov 14, 2019 2:56 pm Your handy cut out and keep guide:

AVENGER
Fullly programmable synth, moderate CPU useage, some sampling and sample-playback abilities.

NEXUS 2 (and 3?)
Preset-based ROMpler, basic tweaks, low CPU usage, no sampling, enhanced multisample playback abilities.

Post

Regardless of whether it's iLok or eLicenser, I'm just glad it's not using that stuff. Because of their decision to ditch that, they have made a sale to me. Will pick it up after Christmas.

Post

mitchiemasha wrote: Thu Nov 14, 2019 11:06 pm
m-ac wrote: Thu Nov 14, 2019 9:53 pm I don't quite follow tbh. Could you give me a simple step by step example of how I could replicate your issues on my own system using nexus?
Actually, there's an easier way, find a simple preset in Nexus, turn everything extra off, duplicate the track, have them both play at the same time. LISTEN! You can even flip the polarity, do they null? This wouldn't work if the sample wasn't using a retrigger mode or using a random phase start position.

Use bpm's other than 60 120 180, due to maths there shouldn't be any rounding at those number to cause the shifts.

The issue is the unintended swing is random and different between each instance.
I tried to "flip the polarity" on 2 duplicate tracks with Nexus 2, Serum and Omnisphere and no plugin were exactly the same, so no they didn't null. I think it's just a normal behavior of all plugins, especially with FX like reverb and delay. The only way I could make them null was to bounce the region in place and copy it twice on two separate tracks, but that was to be expected. Even bouncing the same region twice didn't null with the polarity flipped.

Honestly, I don't see this as a problem at all. Can you give me an audio example of where this could be a problem? I'd be curious to hear it.

Post

m-ac wrote: Thu Nov 14, 2019 8:43 pm
steve.lindqvist wrote: Thu Nov 14, 2019 8:09 pm Image
I'm liking where this is going, refx, you sexy beast!
Nice, they keep giving more details every few days! Can't wait to see a preview of the interface.

Post

Loudness_Contour wrote: Fri Nov 15, 2019 2:45 am
m-ac wrote: Thu Nov 14, 2019 8:43 pm
steve.lindqvist wrote: Thu Nov 14, 2019 8:09 pm Image
I'm liking where this is going, refx, you sexy beast!
Nice, they keep giving more details every few days! Can't wait to see a preview of the interface.
I can't wait to see the browser. I bet they're saving the best for last.

Post

Loudness_Contour wrote: Fri Nov 15, 2019 2:41 am
mitchiemasha wrote: Thu Nov 14, 2019 11:06 pm
turn everything extra off
I think it's just a normal behavior of all plugins, especially with FX like reverb and delay.

Honestly, I don't see this as a problem at all. Can you give me an audio example of where this could be a problem? I'd be curious to hear it.
No it's not an issue for all synths, AVENGER doesn't do it (although it did have a consistent latency delay issue I pointed out in the Avenger page here), Sylenth 1 doesn't do it, Kontakt doesn't do it, Cubase sampler doesn't do it. I specified "turn everything extra off" of course FX will not null. You should be able to hear it everytime you layer 2 instances of Nexus with exact same midi, shifts in phase. If you haven't noticed it yet, you will eventually. I'm not uploading a audio example but I can do one better... This is a direct quote from Michael, the owner of Nexus.

ME: "I've been told that this is because Nexus isn't sample accurate".
Michael: "By whom? Albeit technically correct, the variation is in the sub-millisecond range".

It's only an issue if you want waveforms from different instances of Nexus to maintain perfect phase relationship. Which is very important when layering certain sounds.

Download Shaper Box from Cable Guys and start using it's visualiser.
MSI z390, i7 9700k OC, Noctua Cooling, NVMe 970 Pro, 64GB 3200C16, BeQuiet PSU, W10, Cubase 10, Avenger, Spire, Nexus, iZotope, Virus TI (INTERGRATED), Yamaha CS2x (Custom Mapped).

Post

m-ac wrote: Thu Nov 14, 2019 9:53 pm I don't quite follow tbh. Could you give me a simple step by step example of how I could replicate your issues on my own system using nexus?
Loudness_Contour wrote: Fri Nov 15, 2019 2:41 am Honestly, I don't see this as a problem at all. Can you give me an audio example of where this could be a problem? I'd be curious to hear it.
To follow mitchiemasha’s issue you have to understand the core of the problem. To keep it simple: Nexus is not sample accurate, it means notes aren’t generated on a musical grid but on a ms base. So notes aren’t audible at the position were you play or paint them in your DAW, they shift/drift within a 1 ms time slot. This leads to phasing, timing and transient issues as mitchiemasha's encounters them when layering multiple Nexus instances. But the core of the problem also affects every Nexus user when bouncing a single note in your project - it won't be on time/on your grid.

Here is an explanation from another topic
The audio data is transferred to the DAW depending on the buffer size / samples - so that it does not match the musical grid(!). Why is that? It's just buffer time, so that the plug-in hast time to calculate the sound and doesn't hit the CPU as much (cp. buffer size with your audio interface vs latency vs CPU hit).

Nexus is running on a 1 ms timer! Try tempo 120 Bpm and place your note on every beat - the notes should perfectly in time / on grid as the 1ms timer is matching the musical grid at some Bpm times. You just have to use "full ms numbers" without decimals to get Nexus aligned with the grid (at least in Logic). 60.000 ms (because 60 seconds in a minute is 1.000 ms in a second equals 60.000ms per minute) / 500 ms (full ms number without decimals) = 120 BPM.

The other way round: 60.000 ms / 130 BPM = 461,538 ms -> decimals therefore Nexus isn't aligned to the musical grid on 130 BPM, so you get time shifts. Nexus is only in time if you get the 60.000 ms and your BPM number to full ms number.

Another example: 60.000 ms / 100 BPM = 600 ms -> full ms number therefore Nexus is aligned to the musical grid on 100 BPM.

So the audio from nexus isn't really random it just rounds up the audio material to the next full ms. That's all.

This is an issue several virtual instruments have (as you already find out with the Korg M1, there are several more...). Unfortunately this can't be fixed at ease. I did have discussions with several developers as well as reFX but they don't intend to move away from the 1ms timer. So you have to live with it! A quote from them: Nexus isn't intended to be a drum computer. Unfortunately this is very sad as Nexus has quite good drum samples buried in their SQs. I also do resample the Nexus drums (what irony for such an expensive tool) and use them as audio in the DAW or even in other virtual sample players with perfect timing. Otherwise you can't even get a simple four on the floor kick perfectly on grid (depending on the BPM).

Post

You can easily experience the timing issue yourself with the Nexus init sound within the DAW of your choice:

Open your DAW, set the tempo to 130 BPM.
Open a new instance of Nexus, just use the LD Saw Init preset.
Deactive FX (Delay, Reveb)
Create a two bar part/clip/region
Now paint in 8 short notes (quavers/eighth notes) just one on every beat (simple four on the floor pattern):


Image

Next bounce/render the two bar part/clip/region to a new audio track.
Cut up the part/clip/region on every beat (do not use zero crossing functions), so that you’ll have all the bounced/rendere notes as single slices:


Image

Now create 7 additional audio tracks and drag all the single slices on the first beat on every track, so that you can compare them on top of each other:

Image

Zoom in. Now you can see the inaccurate timing quite well:

Image

Now try this again in different tempi and with different virtual instruments for comparison.

The problem is that these shifts vary from instance to instance and from tempo to tempo because the ms grid does not fit the musical grid. Or in other words, it is hardly possible to play a simple four on the floor kick on point in Nexus.

This is how it should look and sound like:

Image

That's why I'm pleading for a sample-accurate offline export (by sample accurate I mean correctly attached to the musical grid).

Post

do you make music or do a math exam? Honestly... 1ms resolution, thats 1000 steps per second should be way enough if you make MUSIC. To raise the resolution would mean heavy impact on the CPU, and I mean HEAVY. We had a high resolution once in Avenger, but it was unusable, after few voices it got far beyond 100% CPU. So I really never noticed any problems in my 10 years making audio demos and sounds from Nexus. Listen to the audio demos, do you hear a timing problem there?

Post

For 1/1000th of a second - absolutely, I'll take the low CPU over the timing any day.
http://www.guyrowland.co.uk
http://www.sound-on-screen.com
W10, i7 7820X, 64gb RAM, RME Babyface, 1050ti, PT 2023 Ultimate, Cubase Pro 13
Macbook Air M2 OSX 10.15

Post

actually the guy above was asking about OFFLINE accuracy for rendering purposes. So the question is - if it possible to implement hi-res mode for rendering.

Post

msvs wrote: Fri Nov 15, 2019 8:46 am do you make music or do a math exam? Honestly... 1ms resolution, thats 1000 steps per second should be way enough if you make MUSIC. To raise the resolution would mean heavy impact on the CPU, and I mean HEAVY. We had a high resolution once in Avenger, but it was unusable, after few voices it got far beyond 100% CPU.
I fully understand, that's why I think the idea of sample accurate offline bouncing could be an option. Just like other manufacturers do it with the offline oversampling settings (e.g. FXpansion).
msvs wrote: Fri Nov 15, 2019 8:46 am So I really never noticed any problems in my 10 years making audio demos and sounds from Nexus. Listen to the audio demos, do you hear a timing problem there?
In fact, I've never looked for anything like it. But actually yes, I've heard that there's something wrong with the timing (sometimes). Only then(!) did I start to see what might be the reason.

Can I hear one ms differences alone? No, of course not. But as soon as other instruments and tracks are involved (see mitchiemasha’s layering problem) you can hear it. Especially if you have a comparison between accurate and inaccurate timing (e.g. exchanging your Nexus drums against other virtual instruments, instant tightness). The problem is also that there is no uniform shift, this could even be compensated for because it would be consistent in itself. The problem is that every note (depending on the tempo) sometimes comes earlier, sometimes later and that makes no sense musically and leads to phase problems and eats up your transients when cutting bounced audio on a musical grid later on. The notes are moving, I don’t want that usually on my main drums and basses for example.

Anyway, it’s similar to aliasing. Many people don't hear it, but once you have it in your ear, you can't make it unheard. The same is true with timing, if you are used to tight timing (or better transient/impulse tightness) and can recognize it, then you feel immediately if something is wrong.

I assume that you can at least hear a 10 ms difference. You can try these blind tests here:

https://www.audiocheck.net/blindtests_t ... hp?time=10

At „Files being tested“ you can listen and train the two files for a moment and at „The Test“ you can blind test yourself. With 10 ms even the most unmusical person should be able to do it. On top of the site you can go lower to 5 ms, to 2 ms and finally to 1 ms.
Of course you need a good ear and even more good headphones (real, high-quality ones, with very fast transient speed/response).
However, this is only a super super simple example where you might say that's bearable and hard to hear. Don’t forget this test is a „fixed shift“! Nexus is drifting depending on the tempo!
So, if you have several tracks that depend on precision (i.e. no pad, strings, slow stuff etc.), then you have these shifts between all tracks, instances, and notes, and your timing and sound(!) simply starts to get mushy. In addition, the majority (almost all) of current software synthesizer/samplers have a sample-accurate timing.

For someone who has this issue in his ear, a 1 ms shift sounds quite similar like a 10 ms shift to the untrained ear. And don't tell me it wouldn't bother you if the hi-hat would sometimes sit on the kick and sometimes not… (as in the 10 ms example, if you wouldn’t want it explicitly).

Is that why I can't work with Nexus? No, of course not!

Does it bother me? Yes, it annoys me, especially because I would like to use main drum and rhythmical elements (I know it’s not supposed to be a drum sampler) without having to resample/bounce them all the time.

Do I recommend Nexus anyway and praise it?
Hell, yes!

Post

steve.lindqvist wrote: Fri Nov 15, 2019 8:07 am Cut up the part/clip/region on every beat (do not use zero crossing functions), so that you’ll have all the bounced/rendere notes as single slices:[/i]

Zoom in. Now you can see the inaccurate timing quite well:
Might as well be caused by the cutting.

Locked

Return to “Instruments”