FXpansion Cypher2 - Coming in 2018

[Angus_FX]

I'll post some nastier, more aggressive, more nonlinear demos. There are dark, light and high-contrast skins

The GUI is subject to some fundamental constraints:

- Can't be any bigger at its default size, dictated by laptop screens & the need for dock/taskbar and miscellaneous host "window furniture".

- Text can't be any smaller, dicated by legibility & pixel size.

- Can't put individual synth features like the envelopes / LFOs / oscillators on banks/tabs, as that's just a horrid way to work. All the important stuff needs to be on the main page at once, so you don't have to go hunting around too much to figure out how a patch works.

- It's vector-drawn at runtime, which means photo-realistic is impractical; TransMod doesn't have a direct analogy in hardware, so at best it'd be more hyperreal than photoreal. In an ideal world we'd render in realtime 3D, DirectX on GPU, everything with floating-point coordinate positions.. the possibilities with that are amazing especially on a Retina screen (freely zoomable with no loss in quality at all) but we're not quite there yet.

- The knobs can't be any smaller as you have to be able to click the outer & inner parts separately.

There will be a few skins, at least 3. The one below is work in progress, some of the colours aren't right yet, but it should give you an idea of what's possible.

Can you give me an example of a synth with different backgrounds for each area? You want a different colour texture on every panel or..?

[db3]

I'm just as happy with photo realistic or flat look, but in either case the UI should make things easy to understand and inspire. The problem with the new fxpansion plugs is they look like the kind of thing you'd get if a programmer did the design, rather than someone proficient in designing UIs. Serum is an example of how good a modern UI can look and function.

[toothnclaw]

Will v2 sound the same as v1 or not? If not, how close?

[Angus_FX]

@db3 Serum's great, but there are a couple of key differences that make their job rather easier. Firstly, it's a tabbed design, with fewer than half the number of controls on-screen. Cypher isn't tabbed, for a good reason: it's easy to make a clean-looking UI at that size if you have tabs, but it means when you load up a sound you can't see all the shapes of envelopes, LFOs etc. at once, nor reach them for editing with a single click.

The other big one is that not everything on Serum is TransMod, so he can do stuff like those big, nice-looking graphical spline envelopes (FabFilter are another company who do that really well, IMO, but with a vector UI). Whereas due to the nature of Cypher, pretty much everything has to be represented by a knob or slider. In any case, without tabs, there isn't the pixel budget for big interactive visualisers.

As far as I can tell Serum's a bitmapped UI, not resizable, and to my eyes, the labels are honestly a tiny bit fuzzy and hard to read.. they're nicely styled, but it's a soft, chunky font at a fairly small point size.. a high-DPI version for Retina Display would look fantastic, but at standard pixel density it doesn't quite do it for me. Because of the above, synths like Kontor, ARP2600v & Bazille are probably better comparisons in terms of what's practical (and of those three, Bazille is the only one AFAIK with a resizable UI).

@toothnclaw very close indeed, although v2 only runs at 2x oversampled and above. There are some pretty subtle differences in the filters, shapers and exponential envelopes. Happy to post A/B's of specific patches if you're interested.

[toothnclaw]

Thanks, Angus.

[Robmobius]

Cant wait for this - the UI looks slick.

[aMUSEd]

I trust this is just a draft gui - everything is horribly cramped. I like the filter layout in V1 particularly.

[Angus_FX]

General feeling around the v1 layout was although it's superficially easier on the eye, it's hard to actually read ("where's control XYZ?"). I find the same with the Alesis Andromeda.. lovely sounding synth, but the panel layout is really hard to navigate visually because of the curves between the knobs etc.

Btw, when you do Amber 2, include keyboard tracking, but one that allows us to change the nature of it anywhere between linear (100%) and curved, i.e. more than 100%.

Sorry, neglected to answer this earlier. Do you mean keyboard-to-pitch tracking, or filter keyboard tracking, or..? Do you know of any top-octave-divide string synths which do this? Nonlinear pitch tracking is trivial on Cypher2/Strobe2, but on Amber it's harder as all the oscillators are derived from a single master oscillator per note of the scale.

[HunterKiller]

Do we really need photorealistic guis these days? I've always thought of that as a poor attempt to make people think they're not using software. Embrace the future, I say.

That's exactly the problem, that's the wrong way to interpret photorealistic guis. Not that their look has to immitate a hardware box in the first place. And btw, the future is anything you want it to be.

[4damind]

I agree with this photorealistic GUIs. This was good for the first days of virtual instruments to give the user the impression of a real instrument.
For the future I hope for better readable/scalable GUIs and intelligent GUI features and IMO FXpansion is on a good way for this (Afaik Strobe2 is not completely free scalable, I cannot remember about a size drag handle on the GUI?).

[Angus_FX]

It's not free scalable, because for the time being we still have to use integer positioning of controls etc. Rather we offer a choice of ten sizes from 75% to 220%.

Once most people have 4K displays, we can go floating-point with everything (as pixel snapping, pixel-based lines etc. no longer matter at 150ppi).

[HunterKiller]

- Can't put individual synth features like the envelopes / LFOs / oscillators on banks/tabs, as that's just a horrid way to work. All the important stuff needs to be on the main page at once, so you don't have to go hunting around too much to figure out how a patch works.

Agree. But sorting them according to coloured panels, makes the flow of working more logical. Our brains utilize memory more so in pictures than in diagrams. So the more you can create each area of the synth to have its own specific look and environment, the more we are able to identify this quickly and act, not searching for where things are.

- It's vector-drawn at runtime, which means photo-realistic is impractical; TransMod doesn't have a direct analogy in hardware, so at best it'd be more hyperreal than photoreal. In an ideal world we'd render in realtime 3D, DirectX on GPU, everything with floating-point coordinate positions.. the possibilities with that are amazing especially on a Retina screen (freely zoomable with no loss in quality at all) but we're not quite there yet.

Then colour, shade and over all set up, is used to give each area its own unique "personality". I mean, there's more than enough to get inspired by. Some other devs have done things very well.

Can you give me an example of a synth with different backgrounds for each area? You want a different colour texture on every panel or..?

The new Diversion look. Zebra also has different colours for its various modules.
More colour allows you to express the different areas of the synth better. So that it registers better as we process the image on our screen.
http://www.kvraudio.com/forum/viewtopic.php?p=4497084

I agree with this photorealistic GUIs. This was good for the first days of virtual instruments to give the user the impression of a real instrument.
For the future I hope for better readable/scalable GUIs and intelligent GUI features and IMO FXpansion is on a good way for this...

Well not really... Diva, Monark, Vacuum Pro, Olga, OP-X Pro, are all great synths that directly contradict that point of view.
And one can read them too. Scalablity doesn't necessarily mean better, it's just a choice. What is better is what looks, sounds and feels better, not necessarily what only functions better gui-wise. If one can integrate all four, all the better. It's a balance, not a focus on only one aspect of something, anything...

I'm just as happy with photo realistic or flat look, but in either case the UI should make things easy to understand and inspire. The problem with the new fxpansion plugs is they look like the kind of thing you'd get if a programmer did the design, rather than someone proficient in designing UIs. Serum is an example of how good a modern UI can look and function.

Exactly. I mean all of this advice is to make these things better. Resizable uis are nice, but not something extremely important and necessary. Why not have 3, or 4 or 5 different sizes and change between them...?
Gui size has never been a problem for me, it's always the actual look and the way it logically works. If it's set out confusingly, the synth is dead. If you have to convince someone to learn a new way of doing things or a new way of seeing things, (and yet the synth is offering exactly the same character of sounds and a 100 other synths) that's already a challenge most people don't want to undertake, or don't have the time to. On top of the situation we already have where we are swamped with good plugins. It'll sink in the eyes of most, bar a few. Unless it's friendly to use and inspires me. That begins with something beautiful to look at.
Combine that aspect with a good sounds engine and you have a much better chance.

I trust this is just a draft gui - everything is horribly cramped. I like the filter layout in V1 particularly.

Yup, it looks like PolyAna a little bit, just a load of knobs. It makes me think: Who cares! Next!....
It needs a gui stylistic overhaul. I'd hate for them to put all this work in, into version 2 no less, and still have the synth end up being a sleeper. That's discouraging for anyone.

Sorry, neglected to answer this earlier. Do you mean keyboard-to-pitch tracking, or filter keyboard tracking, or..? Do you know of any top-octave-divide string synths which do this? Nonlinear pitch tracking is trivial on Cypher2/Strobe2, but on Amber it's harder as all the oscillators are derived from a single master oscillator per note of the scale.

Filter keyboard tracking, so you can make lower notes slightly darker, imitating the way real world instruments work. It works especially well with synthesizer strings, if you compare the original stringsynth inside the Eminent 310 organ, it has a much more refined sound to it than the very same technology they sold to ARP, which became the Solina SE. One part of that, is that the ARP SSE didn't utilize filters per note, which made each note have its own character, which in turn made the lower notes sounds ever so slightly darker. Which in turn made all the difference perceptually when playing the strings. In comparison the ARP sounds much more raspy and dirty, more a rock and pop instrument, less electronic and ambient as the Eminent 310's strings were. JM Jarre's music for example used the Eminent's more sophisticated sound to create a more subtle sonic pallet for his albums.

The reason for this is that the raspiness - the top frequencies of the lower notes drown out the clarity of the top notes when one is playing a chord in the lower notes and melody with the higher ones. This clarity comes to be appreciated much more when one is able to make the lower notes slightly darker in tone, so that they are more pad-like (which is obviously adjustable) and the top notes come through clearly cutting through and delivering the melody without something constantly buzzing in the background.
I have an quick and dirty example below, comparing the two.

" Do you know of any top-octave-divide string synths which do this? "

No, as was explained above, unless you create a synth in which every note has its own filter settings, as they used to do in the old organs of that era. Because this costs quite a bit to implement, it's not done in hardware nowadays, (and is an all but forgotten art in the software world) but it does give a superior sound, that is: more 3 dimensional, less straightforward and flat. Every hardware instrument used to have its own extra adjustments to the timbre it was outputting (on top of the different components and design), but nowadays this is not done, which is why everything sounds much more similar. It sounds clean!

Here's 2 roughly made comparisons:

1: ARP Solina http://www.mediafire.com/download/aydql ... spy%29.wav
2: Eminent http://www.mediafire.com/download/89wcc ... her%29.wav

Btw, Nightflight has keyboard tracking implemented, but it's linear, so it cuts off too much at the top, but not enough on the bottom. What we want is for the very top notes to not be touched, and for all the notes below that to be gradually, note per note, filtered off as we go down the keyboard, lower and lower in pitch. And to for the user to be able to adjust this effect from overt to subtle, to nothing, so that everyone can get the kind of sound they want and need. It's great to play a string patch where C1 and C2 are not as bright as C5 and C6. When it is uniformly bright across the keyboard, it makes the overall sound you are playing, less classy, less elite, less subtle, less polished, less refined, more uncontrolled, more aggressive, more dirty, more raw.

If you can give the end user the ability to adjust between these two extremes, we can have the best of both worlds, because in some instances, I DO like aggressive and dirty from my strings.

[Angus_FX]

Uh, both those Mediafire links go to the same place.

I was already looking at doing a transistor amp model per key for Amber2, I don't think a simple passive LPF per key will add that much more CPU expense.

[HunterKiller]

Uh, both those Mediafire links go to the same place.

I was already looking at doing a transistor amp model per key for Amber2, I don't think a simple passive LPF per key will add that much more CPU expense.

Typical mediafire screw up, happened before. Let me fix them.

If you can do it properly btw, that'd be a first!

But, some notes require more than just an LPF, some require HPF and Notch as well, that's where it gets tricky, adjusting all that.

[Angus_FX]

Got any references..? HPF isn't any harder than LP, but more than one filter per key starts to get expensive. We'd need, I think, to keep the topology the same across the keyboard even if the RC & tap values can be unique per note.