KNIFONIUM synth released

[Ploki]

I have to ask again: What exactly do you mean by "filter knob lag"? I just tried, and I didn't notice anything remotely like that.

Regarding the aliasing: I don't know really. There's a lot of harmonics produced by the amp. Check a pure sine wave with SPAN. Harmonic-a-licious.

Sort of like Massive X. Even though Knifonium adds a lot more saturation, as it seems.

filter lag:
https://imgur.com/a/Nksm0M1

notice the blip absent in 48k version.

Pure sine wave and triangle are fine really - saw and square aren't

edit
here's the same (but not really the same because pitch-bend acts differently in different sampling rates) note in both 48k vs 192k
https://i.imgur.com/ezft6uS.png
(Re: aliasing)

[chk071]

filter lag:
https://imgur.com/a/Nksm0M1

notice the blip absent in 48k version.

I'm a simple guy... help me a bit out. What exactly did you do there? Automate the filter? Used envelope? Wha?

Note that there are difference in voices when you activate TMT. Filter cutoff changes, resonance changes, envelope stages differ... etc.

[Ploki]

filter lag:
https://imgur.com/a/Nksm0M1

notice the blip absent in 48k version.

I'm a simple guy... help me a bit out. What exactly did you do there? Automate the filter? Used envelope? Wha?

Note that there are difference in voices when you activate TMT. Filter cutoff changes, resonance changes, envelope stages differ... etc.

i automated the filter. compare the two images.

I used the same project, same patch, just switched sampling rate from 48k to 192k.

notice how the steep filter automatization doesn't happen in 48k? because the filter lags too much to even catch it.
at 192k its visible. because this synth acts wierd as hell honestly

[Niowiad]

Why can't someone check this with Plugin Doctor?
Should not be too difficult to solve this question.
Screenshots prefered.

plugindoctor cant process instruments

Oh, my fault.
Developers, here is demand for a new plugin!

A simple spectrum analyzer like SPAN does the job (if we're talking about aliasing).

So, where are the proof screenshots?

I'll say again I still want to buy Knifonium and other PA plugins, and I seriously didn't even want to get into this aliasing argument, but since you guys seem really keen on the matter...

These are both "Init" sawtoot level-matched patches from Repro and Knifonium.
In red you see Repro which shows the usual harmonic content of a sawtooth.
In white you see the Knifonium sawtooth showing a lot of unexpected peaks inbetween the normal even/odd series (showed in red by Repro), not to mention below the fundamental.
You're not going to see those inbetween peaks on lower ranges (like the screenshot posted by chk071) because it's too far from Nyquist, the issue stands on the upper range.



Funny thing with aliasing, it is often said to just use your ears and don't care about spectrum analyzers... don't worry about a graph unless you can actually hear it. And I agree, but this is one of the cases where it's indeed possible not just to see but to even hear it by playing a simple "Init" patch in the upper range.
It's noticeable in the audio samples which Ploki posted earlier, and the fact increasing the sample rate reduces the artifacts is very indicative.

But here's a new example...
Repro > Knif > Repro > Knif ==> https://vocaroo.com/5pLyAWz0R8P
The amount of inharmonic content due to harmonic fold-back in the Knifonium parts is impossible not to notice, it almost sounds like oscillator-sync during the pitchbends.

In any other softsynth of mine I tried it doesn't get nearly as bad, if I have to be completely honest.
I still want to buy Knifonium for lower register stuff, but it's clear at this point that it can't exactly be pushed in the upper register unless you run your sessions at very high sample rates like 192 kHz, which is highly impractical for most people.

[chk071]

So why should sawtooth and square alias but not sine and trinagle? That doesn't make much sense.

[Calenberger]

Why can't someone check this with Plugin Doctor?
Should not be too difficult to solve this question.
Screenshots prefered.

plugindoctor cant process instruments

Oh, my fault.
Developers, here is demand for a new plugin!

A simple spectrum analyzer like SPAN does the job (if we're talking about aliasing).

So, where are the proof screenshots?

I'll say again I still want to buy Knifonium and other PA plugins, and I seriously didn't even want to get into this aliasing argument, but since you guys seem really keen on the matter...

These are both "Init" sawtoot level-matched patches from Repro and Knifonium.
In red you see Repro which shows the usual harmonic content of a sawtooth.
In white you see the Knifonium sawtooth showing a lot of unexpected peaks inbetween the normal even/odd series (showed in red by Repro), not to mention below the fundamental.
You're not going to see those inbetween peaks on lower ranges (like the screenshot posted by chk071) because it's too far from Nyquist, the issue stands on the upper range.



Funny thing with aliasing, it is often said to just use your ears and don't care about spectrum analyzers... don't worry about a graph unless you can actually hear it. And I agree, but this is one of the cases where it's indeed possible not just to see but to even hear it by playing a simple "Init" patch in the upper range.
It's noticeable in the audio samples which Ploki posted earlier, and the fact increasing the sample rate reduces the artifacts is very indicative.

But here's a new example...
Repro > Knif > Repro > Knif ==> https://vocaroo.com/5pLyAWz0R8P
The amount of inharmonic content due to harmonic fold-back in the Knifonium parts is impossible not to notice, it almost sounds like oscillator-sync during the pitchbends.

In any other softsynth of mine I tried it doesn't get nearly as bad, if I have to be completely honest.
I still want to buy Knifonium for lower register stuff, but it's clear at this point that it can't exactly be pushed in the upper register unless you run your sessions at very high sample rates like 192 kHz, which is highly impractical for most people.

IMO the pattern is too regular to be aliasing, I think it's some kind of harmonising.

[Ploki]

It has distinct counter motion.

Also, parameter lag is real. Which is more concerning than aliasing...

@chk071 because saw and square probably alias at oscilator level, not from saturation

[Niowiad]

So why should sawtooth and square alias but not sine and trinagle? That doesn't make much sense.

It actually makes a lot of sense because the saws and pulses are much richer in harmonic content, meaning that the level of upper harmonics in saws and squares is much higher than triangle and a sine.
Sine ideally has no upper harmonics, but the Knifonium sinewave is saturated to the point of almost being a triangle.

This are the four shapes in Knifonium



The higher/louder the upper armonics are, the higher/louder they're going to be when folded back at Nyquist.

[Calenberger]

Obviously a very special synth with a very special sound signature.
Good? Bad? Don't know.

[Ploki]

Parameter lag changing with sampling rate is not special - just malfunctioning

[Niowiad]

IMO the pattern is too regular to be aliasing, I think it's some kind of harmonising.

They visually showed contrary motion when pitchbent, as said by Ploki.
And in any case, what matters is the audio, which gets as in-harmonic as it can possibly do.
https://vocaroo.com/5pLyAWz0R8P

[Calenberger]

IMO the pattern is too regular to be aliasing, I think it's some kind of harmonising.

They visually showed contrary motion when pitchbent, as said by Ploki.
And in any case, what matters is the audio, which gets as in-harmonic as it can possibly do.

But that's not aliasing, something else, right?

[Ploki]

Contrary motion of inharmonic content is distinctively aliasing.

[Niowiad]

IMO the pattern is too regular to be aliasing, I think it's some kind of harmonising.

They visually showed contrary motion when pitchbent, as said by Ploki.
And in any case, what matters is the audio, which gets as in-harmonic as it can possibly do.

But that's not aliasing, something else, right?

If that's not aliasing, I have no idea what it is, quite frankly.
Putting a non-oversampled distortion plugin on top of Repro gave similar results.

[Calenberger]

I've seen some graphs of aliasing plugins, their pattern were completely different, not that regular.