FXpansion releases Cypher2

[EvilDragon]

Trying out Cypher beta. Seems that there are still some weirdities with transmod...

So let's say I modulate filter cutoff with a bipolar LFO. Cutoff is at 12 o'clock and I put a fair amount of LFO modulation in, so that it basically covers the whole knob. Now if I start moving the cutoff parameter itself to maximum value, I notice how the modulation amount actually decreases! It doesn't go to zero, but it definitely doesn't have the full swing it used to have. I know of no other synth that behaves like this w.r.t. modulation amounts. Here, a GIF should help (I am holding one note at all times here):



This design makes zero sense to me. Any other synth, the negative part of LFO would still act in full swing with things set up like this. Why does transmod reduce the mod amount here? At least the modulation doesn't vanish like it used to, but this is still what I would call unexpected behavior.


I also have a weird issue with the mouse, where I cannot seem to turn the knob fully in one mouse drag movement - it just stops dead randomly and forces me to click and drag again. What the hell?

[Echoes in the Attic]

Trying out Cypher beta. Seems that there are still some weirdities with transmod...

So let's say I modulate filter cutoff with a bipolar LFO. Cutoff is at 12 o'clock and I put a fair amount of LFO modulation in, so that it basically covers the whole knob. Now if I start moving the cutoff parameter itself to maximum value, I notice how the modulation amount actually decreases! It doesn't go to zero, but it definitely doesn't have the full swing it used to have. I know of no other synth that behaves like this w.r.t. modulation amounts. Here, a GIF should help (I am holding one note at all times here):



This design makes zero sense to me. Any other synth, the negative part of LFO would still act in full swing with things set up like this. Why does transmod reduce the mod amount here? At least the modulation doesn't vanish like it used to, but this is still what I would call unexpected behavior.


I also have a weird issue with the mouse, where I cannot seem to turn the knob fully in one mouse drag movement - it just stops dead randomly and forces me to click and drag again. What the hell?

Yeah that's not right.

[Angus_FX]

This design makes zero sense to me. Any other synth, the negative part of LFO would still act in full swing with things set up like this. Why does transmod reduce the mod amount here? At least the modulation doesn't vanish like it used to, but this is still what I would call unexpected behavior.

The issue is that the mod *depth* itself is still a control signal, and all control signals are limited so as not to exceed the max allowed value for the parameter. For unipolar sources the behaviour is more or less as you'd expect (but the _amplitude_ (or depth) of the modulation becomes limited, rather than the maximum value hard-clipping at max-val). For bipolar sources, I'll grant you that it's a little counterintuitive. But I think better to make this fix available as-is than to wait for another QA window (we've a small team working on a lot of projects).

Note that the depth originally dragged by the user is preserved, so that when you dial back to a lower main-knob value, the range is restored to whatever you originally set.

[EvilDragon]

Yes I did notice it's preserved when you drag it back down. But this is still really wrong behavior. It should be clipped at min/max param values when actually applied in the engine, but it shouldn't rescale the other side of bipolar modulation source, ever...

I agree that fixing the bug where you completely lose the modulation when moving the knob to max or min is better to be released now. But I am really hoping the fix for this unexpected/counterintuitive behavior is not gonna be a long wait either... (Also I'd argue that "a little counterintuitive" is putting it extremely mildly, when no other synth behaves this way, as far as I can tell!)

Angus, can you confirm the mouse issue I seem to have? I start tweaking a parameter (say, move filter resonance from 0% to 100%), and Cypher/Strobe don't really allow me to do it in one go straight away, instead I get blocked along the way, then I have to drag again... really weird. Had the mouse cursor been visible during this operation it'd make more sense to show it in a GIF, but I hope the words convey it well enough instead.

EDIT: Here's a GIF anyways:



Every time you see the cursor disappear, I am basically clicking left mouse button and sliding the mouse all the way up in effort to reach the maximum knob value. It never really happens in one go, I either have to release click then try again, or return back to minimum and drag back up to full value again.

What's happening here? This is making your synth really hard to use, I constantly have to fight the knobs.

[Angus_FX]

I'm not seeing that, no. The only time it does anything similar for me is if I move the cursor really slowly and it reaches the edge of the screen before the parameter hits max.

What screen rez, magnification, mouse sensitivity level etc. are you running?

[EvilDragon]

1920x1200, no magnification (100% OS-level scaling in Windows 10), 135% zoom level in Cypher/Strobe, mouse sensitivity in Windows set to default (in the middle). Logitech MX Master 2S mouse.

Playing the devil's advocate: if the cursor is being hidden (which it is in your synths), shouldn't the cursor not really be able to hit the edge of the screen when hidden - you basically calculate deltas and keep the cursor in the same location until it reappears?


EDIT: OK yeah. This totally depends on where the plugin window is. If it's close to the top of the screen, and I tweak filter cutoff in Cypher, then it stops at some point. If I move the window down, I can move it full spread.

This is also what I would call an unexpected behavior. I know in Surge when we implemented hiding cursor, we did so as I mentioned above - calculating deltas, cursor never moves, so there's no chance to hit a roadblock like this.

(On that note, there should definitely be an option NOT to hide the mouse cursor, so that users with a Wacom tablet can also use your synth normally. )

[WasteLand]

and i guess touchscreen friendly interface... will be for a next time? or no-time.... (if cypher is made in JUCE, it seems not that problematic..).

strange how, indeed a wacom tablet, a normal mouse (...) and a touchscreen all have different behavior, o well not that strange.. but still in these times?

(with touchscreen i mean, not a digitizer... or surface.. just a monitor, with touch-capabilities).

pity about the mod depth bug, that EvilDragon discovered. it was sensible to have fully working TransMod environment, one of the main selling points (beside the very own character of cypher 2, and MPE..).

[Angus_FX]

Playing the devil's advocate: if the cursor is being hidden (which it is in your synths), shouldn't the cursor not really be able to hit the edge of the screen when hidden - you basically calculate deltas and keep the cursor in the same location until it reappears?

The cursor's hidden rather than pinned, but I take your point.

How is a Wacom tablet different to a trackpad in this respect?

Either way - we're going to push this update out as there's not another QA window for a while and it's considerably better than previous (and has native M1 support which will be important to many), but will keep track of your reports for a future one.

[Angus_FX]

and i guess touchscreen friendly interface... will be for a next time? or no-time.... (if cypher is made in JUCE, it seems not that problematic..).

The problem with touchscreens is that the hitpoints on Transmod controls are too small. Touchscreens need larger controls at lower density. It's why we have an ARM Mac version but no iPad version - the GUI would need a total redesign to be physically usable, the widgets are too small even on an iPad Pro.

[EvilDragon]

How is a Wacom tablet different to a trackpad in this respect?

AFAIK Wacom tablet requires absolute pointer device, which means cursor shouldn't be hidden nor pinned. Probably not that much different from a trackpad, I suppose!

as there's not another QA window for a while

I assume that's at least several months on end, then...

[WasteLand]

and i guess touchscreen friendly interface... will be for a next time? or no-time.... (if cypher is made in JUCE, it seems not that problematic..).

The problem with touchscreens is that the hitpoints on Transmod controls are too small. Touchscreens need larger controls at lower density. It's why we have an ARM Mac version but no iPad version - the GUI would need a total redesign to be physically usable, the widgets are too small even on an iPad Pro.

ok thanks i will try the old approach; mapping it (parameter mapping) which also works.

[Noumena]

so this instrument will continue tonbe broken for how many months before the next qa window? completely bizarre.

[Angus_FX]

AFAIK Wacom tablet requires absolute pointer device, which means cursor shouldn't be hidden nor pinned. Probably not that much different from a trackpad, I suppose!

We don't pin the cursor. Seems to work OK hidden, haven't had any complaint from Wacom users.

so this instrument will continue tonbe broken for how many months before the next qa window? completely bizarre.

It's working well enough for our sound designers & the beta team who've had it for almost a month. It's not a crash, data loss or similar. Indeed, I checked with Synth Squad v1 and it's been that way all the way back to v1.0 in 2009, and the alpha/beta versions before that. So I feel your characterisation of "broken" is inaccurate and wrong. (Try it with a bipolar LFO on Cypher v1 - the behaviour is exactly the same).

Same goes for the cursor "clip" behaviour. The v2 synths hide the cursor but don't pin; SSv1 neither hides nor pins so it's a bit more obvious when cursor clipping happens.

[EvilDragon]

So do you agree that if you're hiding the cursor, it should just be pinned and work on accumulated deltas, rather than what it's doing now? Because frankly all those roadblocks when tweaking certain knobs really make for a very unsatisfactory UX.

The way I see it, there should be an option to hide the cursor when editing parameters or not, and an option to set the cursor to absolute (current) vs relative (pinned) behavior. Whenever you hit that next QA window, of course.

[zvenx]

I can honestly say that has never once bothered me. I didn't even notice it till now and I have been using them from the get go.
rsp