Is this Zebra3 at SuperBooth???

[Urs]

Ok, one more to show that it also actually works:

[edit: check video below]

I haven't implemented a classic phase distortion algorithm yet, but there's a "Rephase" OscFX that remaps the phase of a waveform from another waveform. In order to do classic Phase Distortion I need to add a raised Cosine waveform (3rd waveform, positioned at 100) to the two waveforms that map the phase.

Then, the Oscillator gets positioned to play the raised cosine waveform. The Rephase OscFX gets set to use the curve. The first two waveforms on the curve, which are positioned 0 and 50 get morphed by the OscFX Position knob from 0 to 50 (as are their positions on the "morph timeline").

The output is then classic Phase Distortion.

But of course, we can go further and change the base waveform from raised cosine to any other on the "morph timeline", which might have interesting results, too.

[gentleclockdivider]

Ok, one more to show that it also actually works:



I haven't implemented a classic phase distortion algorithm yet

I assume you are talking about the upcoming zebra(lette) 3 ?
The phase distortion oscfx in zebra 2 Hz work perfectly , except for resonant waveforms , but that wasn't PD in the first place ( windowed sync )

[Urs]

I assume you are talking about the upcoming zebra(lette) 3 ?

Yes. Much of the OscFX are still work in progress, even if the looks of this seem pretty mature. We only started an initial visual design a few weeks ago - only so it wouldn't look too shabby at Superbooth.

[gentleclockdivider]

Can we still use the old skool method ?
When using the phase distortion effect it is essential to merge points (overlap )
See example

Yes, it absolutely still works:

Thanks , but when ( if ) the pd algo is implemented , I think the highlighted point of the square ( red rectagle ) might cause issues when distorting a part of the cosine .
It should be located where the green arrow points to , on the same x-axis as shown in my giff animation )
I just hope it will be compatible because the PD osc efect in zebra is so good , combined with a tiny amount of DX effect

[Urs]

Hah, I missed that. I was only focussed on how the other points would assign automatically.

One simply has to add another point at the beginning of the path:



Good catch! - I had been wondering about the usefulness of vertical lines at the beginning or end of the waveform, and I currently simply optimise them away after most operations. Giving the option to automatically close the waveform at the end or beginning would certainly make this rather easy. I'll think about that a bit more.

[gentleclockdivider]

Awesome !

[Urs]

Yep, "close left" and "close right" seem to do the trick - when one of them is enabled, the editor acts like the one in Zebra2 where selecting the first point automatically also selects the last point. So far this doesn't seem to have any side effects with any of the other tools, yay!

[vertibration]

What does Z3 wave morphing offer that others do not? How does it differ, and what are the benefits over other synths?

I kind of don't understand the question. Which other synths offer geometric morphing of user editable waveforms?

In my view not everything that's called "morphing" is technically similar to the effect in film and music videos which defined the term (Terminator 2, "Black or White" by Michael Jackson). In music applications the term is often used for all sorts of crossfades and interpolation techniques (e.g. "spectral morphing" in Hive, also Serum and Vital IIRC), but these are by no means comparable technologies.

The kind of morphing that I'm talking about is geometric. It requires that a waveform (or shape, or 3D model, if we include 2D/3D applications) is either composed of mathematical curves (Splines, NURBS, Paths), or masked by them. These curves/paths/outlines are then seamlessly transformed from one shape to another while their visual content is geometrically distorted and - in case of 2D images or textures - crossfaded in the process. The defining aspect is the change of shape though, which is achieved by interpolating the geometry rather than crossfading it.

The only synth I recall to use some sort of user editable spline based waveform for its oscillator is Curve by Cableguys, but it does not perform any morphing between two or more shapes. Synths like Vital or Serum let users edit waveforms that are sample based, so there is no spline based outline which lets the user define seamless geometric transitions.

Hence, I would not know which synth to compare this to.

I understand now, thank you for explaining

[Tj Shredder]

This morphing of spline based waveforms should also allow to do Vosim synthesis by using the note/pitch information as modulator… That would make it easy to have formants in the sound without any filters… In general I find filterless synthesis more expressive, especially when played with an expressive controller…

[DrGonzo]

I've been waiting for this this for years, and that's still half the time others have waited for it.
When this things drops, I'm going to go full in on it.

C A N T W A I T

[gentleclockdivider]

This morphing of spline based waveforms should also allow to do Vosim synthesis by using the note/pitch information as modulator… That would make it easy to have formants in the sound without any filters… In general I find filterless synthesis more expressive, especially when played with an expressive controller…

You can already have formants without using filters ,in fact you can fake a sweeping hp-bp-lp filter ( + resonance ) with the spectromorph module .
With Vosim( form of windowed sync ) the windowing function is of utmost importance , just morphing sine waves will not cut it
There is a reason why the the casio cz resonant waveform ( which is windowed sync ) can not be done (accurately ) in zebra
You can esily do it with talmod and reaktor
https://app.box.com/file/721814743584?s ... tccwi12e3m
https://app.box.com/file/721848111303?s ... c6kusfp6x1

[Urs]

So in Z3 it'll be easy to morph both the waveform that is being synced and the window above it. It'll use OscFX to do the sync and apply the window though.

To a degree it's possible to recreate CZ resonant waveforms by morphing waveforms, but it requires an excessive amount of key frames. There's a trick to do it by morphing just two waveforms, but the option behind it is a little more taxing on CPU. I have not followed up on it for that reason.

If someone had a visual example as to what VOSIM does, I'd be happy to try something out.

[padillac]

If I remember correctly, u-he’s assertion for several years has been “people mostly can’t make useful wavetables.” It looks like Z3 intends to change that?

[Urs]

If I remember correctly, u-he’s assertion for several years has been “people mostly can’t make useful wavetables.” It looks like Z3 intends to change that?

indeed.

Although, technically a wavetable is a large bunch of samples, and this is a small bunch of mathematical curves, much like an envelope.

My premises have always been that the most interesting aspect of wavetable scanning are seamless, continuous sweeps in timbre, like in the olden PPG synthesisers. But any built-in wavetable editor based on samples, does not offer the tools to do that (I'll note one exception later). Instead, while users can paint or stitch individual waveforms in series, the means to create smooth transitions between those are merely akin to crossfading between two waveforms. No tool I came across really allows for transitions that look (and sound) like actual morphing in a similar sense as the term is used in motion picture.

A quick example: If in Zebra's "GeoBlend" mode you draw two pulse waves, one 10%, one 90% and you let the software create a transition, what do you get half way in? - It's not a 50% pulse wave, but it is a waveform that looks like a stair case with 2 steps instead of one.

Yes, there are methods to work around this limitation. The wavetable editor from Synthesis Technology for instance lets one gradually apply some kinds of wave shaping, skewing or other processes. A similar thing is possible in Vital, where a number of processes can be keyframed over the size of the wavetable. This is elegant, innovative and great. So now I did see an example that falsifies my original hypothesis to a large degree. But still, the curve based method we had always had in Zebra 2 offers a lot more control over the transition of the shape, and the method mentioned here still does not solve the issue of morphing from one given waveform to another given waveform.

We did work on this problem in Hive's uhm scripting, with mixed results. There, two waveforms could be analysed in spectrum, formants could be extracted, matched between two waveforms and recombined in a seamless transition. While this is exactly the voodoo I was looking for, the settings for this kind of synthesis are hit and miss. I'd spend more time optimising the settings for a good transition than I would spend painting the damn thing myself.

And that is really what it is about. Our curve editor will be fast to work with and results will be predictable and instant. And really, really smooth.

(While again, technically it isn't a wavetable editor, it can of course render the curves into wavetables and thus be used for wavetable synthesis as well.)

[pdxindy]

We did work on this problem in Hive's uhm scripting, with mixed results. There, two waveforms could be analysed in spectrum, formants could be extracted, matched between two waveforms and recombined in a seamless transition. While this is exactly the voodoo I was looking for, the settings for this kind of synthesis are hit and miss. I'd spend more time optimising the settings for a good transition than I would spend painting the damn thing myself.

And that is really what it is about. Our curve editor will be fast to work with and results will be predictable and instant. And really, really smooth.

(While again, technically it isn't a wavetable editor, it can of course render the curves into wavetables and thus be used for wavetable synthesis as well.)

I haven't seen anything like this. As you say it is curve based, not pixel based like wavetable editors. It will be interesting to explore the conversion of samples to curves! Also, the revised OscFX concept. I wonder if one of the new OscFX would give some noisy results?

Very much looking forward to trying this out! What a playground!