[MXXX] More observations and suggestions after spending some time with beta 2

[farfadetfarfelu]

Vojtech, you are probably going to hate me, because i have more observations / suggestions after spending some time with MXXXbeta2 :

- Filter module : add bandpass and notch in the filter module with Q from 0 to 100

- Actually my dream filter module would only have LP 12, HP 12, Bandpass and Notch, each with Q from 0 to 100, and a visual representation of the filter. Not sure all the variety of LP and HP are needed here since they already are in the bandpass module.

- Filter module : Sidechain : addition of an on/off switch to be able to disable the side chain without losing the side chain settings. When sidechain is disabled, the max settings will control the filter Freq and Q.

- Distortion module : input gain control can be useful for subtle drive or as a boost

- Saturator module : input gain increased to 40dB output gain to -40dB for those crazy flat out distorted sounds and more flexibility with lower input signals.

- Granulator : the feedback send does not show when right clicking on the processor in the grid

- Feedback module : add clean path through the feedback module + add mixer (with FM and ring mod) in the feedback processor to mix the clean and feedback together. The way it is now, we have to use a feedback + a mixer module anyway.

- Bandpass module : add a frequency link button to tie the filter frequencies of the LP and HP together. Example, if the HP is set to 100 hz and the LP to 300Hz, when moving the HP from 100 hz to 1100hz, the LP also moves from 300Hz to 1300Hz because they are linked.

- Vibrato, Tremolo, chorus etc rate increased to 100Hz

- LFO modulator : increased max rate (10hz maximum as it is now)

- Chorus : Digitech Multi-Voice Chorus has 16 voices (no idea if 16 voices is useful but i like the completely crazy out there noise that can be done in the 10 voice mode so who knows what more voices might bring...)

- FILTER + EQ : Very high volume peaks (+300dB ? ) occur when using certain modulation settings to modulate a filter frequency.
> For example a follower with a very low release time, or a steep step sequencer, or the filter sidechain on some settings.
> This happens when the Q of the filter is higher than 2 or 3.
This might crash MXXX or even the host. Reaper automatically mutes itself when this happens. Maybe this is a limitation of filter algorythms ? But it's daunting when you are experimenting with the follower and suddenly everything goes silent and you have to reload MXXX or the project. (Mfilter does the same thing).

- Also when using the feedback module with zero or very low latency, it is definitely not a good match with some of the modulators (very high CPU, glitches when adjusting the waveshape of the LFO).
> For example try a feedback module from the output of the distortion processor back into the input of the distortion. Then modulate the distortion « color » with a follower... huge cpu load with low feedback settings.
> other example, just insert a feedback module with zero latency somewhere on the grid, now insert a phaser and try to modulate the rate with an LFO : on my system, CPU jumps to 10% and glitches when adjusting the lfo waveshape and parameters.

It also does this when adjusting waveshapes inside effects (for example the tremolo waveshape).

It does this even if the feedback processor is in the grid but turned of (ctrl + click) and i just had the case when it continued to do it even after i deleted the feedback from the grid.

- Delay Processor : using a modulator (LFO) on the delay time (increased to 10s) seems problematic, big cpu loads.

Sorry for the long post but i think MXXX is worth the effort.

[MeldaProduction]

Vojtech, you are probably going to hate me, because i have more observations / suggestions after spending some time with MXXXbeta2 :

don't worry

- Filter module : add bandpass and notch in the filter module with Q from 0 to 100

Ok

- Actually my dream filter module would only have LP 12, HP 12, Bandpass and Notch, each with Q from 0 to 100, and a visual representation of the filter. Not sure all the variety of LP and HP are needed here since they already are in the bandpass module.

Well, here the variety of filters is not a problem and filter can be modulated by the side-chain, that's the whole purpose. Visual represation rather no at the moment as it doesn't say too much and it would take some unnecessary CPU.

- Filter module : Sidechain : addition of an on/off switch to be able to disable the side chain without losing the side chain settings. When sidechain is disabled, the max settings will control the filter Freq and Q.

Well, no big deal, but what for?

- Distortion module : input gain control can be useful for subtle drive or as a boost

But there is "drive" parameter which is nothing else than input gain . It is for amp simulator only, but in MMultiBandDistortion there is a global input gain and here you can just use MUtility. You have to understand that one of the purpose of MXXX is to minimize feature duplicities, because it takes almost nothing to add such simple modules.

- Saturator module : input gain increased to 40dB output gain to -40dB for those crazy flat out distorted sounds and more flexibility with lower input signals.

Hmmm, well, ok, here it may make sense, though this will limit compatibility with separate plugin presets.

- Granulator : the feedback send does not show when right clicking on the processor in the grid

It doesn't because it has multiple outputs. But I'll check how simple it would be to send the first one.

- Feedback module : add clean path through the feedback module + add mixer (with FM and ring mod) in the feedback processor to mix the clean and feedback together. The way it is now, we have to use a feedback + a mixer module anyway.

Same thing - mixer may potentially get quite complicated in the future and that would make feedback complicated as well, not mentioning the amount of unnecessary work. So please just use the mixer, it will not take almost any more CPU.

- Bandpass module : add a frequency link button to tie the filter frequencies of the LP and HP together. Example, if the HP is set to 100 hz and the LP to 300Hz, when moving the HP from 100 hz to 1100hz, the LP also moves from 300Hz to 1300Hz because they are linked.

2 problems:
1) It must be logarithmic, not linear as you proposed.
2) When you get to the end of the range, the interval won't be maintained anymore.
So there is a simple solution - use a multiparameter . If you don't use interval range mode, it will even solve the second issue .

- Vibrato, Tremolo, chorus etc rate increased to 100Hz

Ok, so be it. Though again - compatibility limitations with separate plugins... but so what...

- LFO modulator : increased max rate (10hz maximum as it is now)

Here it is! Right! There is the "high-speed" switch which is there for backward compatibility of other plugins! Ok, so let's remove it...
Wait, it's now limited to 1Hz with highspeed to 10Hz. There's a reason for it - modulator protection, it probably won't work very well above 10Hz, we'll see.

- Chorus : Digitech Multi-Voice Chorus has 16 voices (no idea if 16 voices is useful but i like the completely crazy out there noise that can be done in the 10 voice mode so who knows what more voices might bring...)

I'll check but as far as I remember, more voices than 10 seemed to have no big impact except for the CPU .

- FILTER + EQ : Very high volume peaks (+300dB ? ) occur when using certain modulation settings to modulate a filter frequency.
> For example a follower with a very low release time, or a steep step sequencer, or the filter sidechain on some settings.
> This happens when the Q of the filter is higher than 2 or 3.
This might crash MXXX or even the host. Reaper automatically mutes itself when this happens. Maybe this is a limitation of filter algorythms ? But it's daunting when you are experimenting with the follower and suddenly everything goes silent and you have to reload MXXX or the project. (Mfilter does the same thing).

Yeah I noticed... damn, seems this is the nature of the filters, will check what can be done.

- Also when using the feedback module with zero or very low latency, it is definitely not a good match with some of the modulators (very high CPU, glitches when adjusting the waveshape of the LFO).
> For example try a feedback module from the output of the distortion processor back into the input of the distortion. Then modulate the distortion « color » with a follower... huge cpu load with low feedback settings.
> other example, just insert a feedback module with zero latency somewhere on the grid, now insert a phaser and try to modulate the rate with an LFO : on my system, CPU jumps to 10% and glitches when adjusting the lfo waveshape and parameters.

It also does this when adjusting waveshapes inside effects (for example the tremolo waveshape).

It does this even if the feedback processor is in the grid but turned of (ctrl + click) and i just had the case when it continued to do it even after i deleted the feedback from the grid.

It does that after you deleted it -> that's a problem! Will be fixed.
But otherwise I'm afraid there is nothing that can be done. When the feedback delay is lower than actual buffer size, it must subdivide the processing buffers and that takes quite some CPU. In fact, it is the same as extreme modulator protection. But it just cannot be solved. Also some parameters need lots of CPU to update their processing states, so you just shouldn't modulate them, at least not so much .

- Delay Processor : using a modulator (LFO) on the delay time (increased to 10s) seems problematic, big cpu loads.

Yes, that again cannot be solved I'm afraid. It must resize the whole buffer and that takes some time. I was experimenting with some antiglitching some time ago, but this seems the only truly functional way.

[farfadetfarfelu]

Thanks for the details.

- Filter module : Sidechain : addition of an on/off switch to be able to disable the side chain without losing the side chain settings. When sidechain is disabled, the max settings will control the filter Freq and Q.

Well, no big deal, but what for?

Well the sidechain and the modulators can make different sounds, so i was comparing the side chain Vs the "modulators". It's simple to turn of a modulator, but it's less simple to turn of the sidechain.
Also sometimes you will want to use a modulator and not the sidechain, because you already have the modulator set up and you want the filter to be synched to some other modulated parameters that are already all patched up.
A sidechain on/off seems like the obvious solution instead of setting up multiparameter and a utility module to kill the sidechain input.

- Delay Processor : using a modulator (LFO) on the delay time (increased to 10s) seems problematic, big cpu loads.

Yes, that again cannot be solved I'm afraid. It must resize the whole buffer and that takes some time. I was experimenting with some antiglitching some time ago, but this seems the only truly functional way.

Okay well in that case i submit two alternative options here :
- add a switch on Delay tap1 and delay tap 2 to chose between the 10 second mode and a shorter delay mode.
- Or Add two more delay taps :
Tap 1 and 2 limited to 500 ms or whatever delay time works well with modulation
Tap 3 and 4 with 10 second delay time, and the delay time parameter does not appear in the modulation parameter list.

Or maybe one second or two seconds instead of ten is enough if it works well with modulation.

[MeldaProduction]

Well the sidechain and the modulators can make different sounds, so i was comparing the side chain Vs the "modulators". It's simple to turn of a modulator, but it's less simple to turn of the sidechain.
Also sometimes you will want to use a modulator and not the sidechain, because you already have the modulator set up and you want the filter to be synched to some other modulated parameters that are already all patched up.
A sidechain on/off seems like the obvious solution instead of setting up
multiparameter and a utility module to kill the sidechain input.

Aaaaah, ok, that makes sense, though the purpose of the filter processor is the sidechain, so you can just use the band pass. But with the more filters and for convenience the filter module shall be better, so ok .

Okay well in that case i submit two alternative options here :
- add a switch on Delay tap1 and delay tap 2 to chose between the 10 second mode and a shorter delay mode.
- Or Add two more delay taps :
Tap 1 and 2 limited to 500 ms or whatever delay time works well with modulation
Tap 3 and 4 with 10 second delay time, and the delay time parameter does not appear in the modulation parameter list.

Or maybe one second or two seconds instead of ten is enough if it works well with modulation.

I don't get it at all. These are just limits, you can use whatever values you want and every modulator/multiparam/controller has a range, so what is it all about??

[farfadetfarfelu]

Okay well in that case i submit two alternative options here :
- add a switch on Delay tap1 and delay tap 2 to chose between the 10 second mode and a shorter delay mode.
- Or Add two more delay taps :
Tap 1 and 2 limited to 500 ms or whatever delay time works well with modulation
Tap 3 and 4 with 10 second delay time, and the delay time parameter does not appear in the modulation parameter list.

Or maybe one second or two seconds instead of ten is enough if it works well with modulation.

I don't get it at all. These are just limits, you can use whatever values you want and every modulator/multiparam/controller has a range, so what is it all about??

Ok I'll explain my view.
On my system, when i try to modulate the delay over a one second range, it's really bad : audio glitches + it slows down the plugin and host response.
You will have users trying to modulate the delay time and have this happening on their system :
- They might get the impression the VST is buggy
- They might get the impression the vst is unreliable.
- They might come asking for support on this forum
So yes it is possible to limit the range of the modulator so that the delay time stays in a non-glitchy range, but when you first hook up the modulator, the range is not fine-tuned so you get this bad impression that's it's not working.
That's why i proposed the alternative suggestions.

Actually it's somewhat the same debate about the low feedback settings, should you allow no latency feedback when it might cause weird behaviours with the plugin for a number of users who might not really understand what's going on.
But this could be solved by a little warning message in the feedback processor. (i just noticed there is already a note about this when clicking F1 on the feedback delay control).

[MeldaProduction]

Hmmm, maybe you are right. On the other hand this would mean more complicated GUI... dunno...

[salatmensch]

hiya,

i've got two other observations. one is the colours. when i use the "gray" colour scheme, some of the help is not really readable on my screen. the yellow and green is far to bright. (have a mac book pro screen)

i find the limiter really confusing. i can't really get the limiter to work - or i just don't understand it at all. it always seems to introduce distortion. it sounds actually worse then not having it in. for example. i get occasional peaks of +4 dbfs. so i put the limiter in and not mess with the settings. then the actual peaks get limited but sound horribly clipped. so i have to manually reduce the input level by 4 db not not have them sound distorted. am i doing something wrong? the effect is even worse once i change the limiter into "logarithmic" mode and set it to "clean". altering the attack/look-a-head-settings also don't really help.

and then i thought i have to set dry/wet setting in the saturation tab to "100%". then i the limiter seems to add some soft clipping which reduces the annoying distortion quite a bit. but looking to the help it tells me that this step is actually something to add extra flavour.


also i noticed that the first rows of settings (input gain, ceiling, threshold, kneesize) doesn't have the help button. not sure if this is meant to be like this. other modules such as the dynamics one have a help button for the "general parameters".

bests,
salat

ps:i should mention that i'm using mxxx in 7.1. mode

[farfadetfarfelu]

When using the limiter with strong input signals, I had acceptable results by keeping a low attack and setting the release time higher (100 ms).
(with the saturation dry/wet left at zero)
I did not notice any peaks slipping by.

With a fair amount of look ahead, If you move the attack slider fast you will notice a weird tape slowing style effect...

[salatmensch]

hiya,

i managed to make the matrix window of the vocoder inaccessable. this is how i managed to do this. first i set up the number of bands to 100 and opened and closed the window. it barely fitted the screen (have a resolution of 1680x1050). then i reduced the number of bands to 32. now i opened the window and i can barely see the ok button. most of the window has disappeared in the virtual realm above the screen. the problem now is that i cannot move the window down because i don't have access to the top of the window. removing mxxx and putting it back in doesn't affect the window position. it will stay beyond my screen size.

bests,
salat

[salatmensch]

When using the limiter with strong input signals, I had acceptable results by keeping a low attack and setting the release time higher (100 ms).
(with the saturation dry/wet left at zero)
I did not notice any peaks slipping by.

With a fair amount of look ahead, If you move the attack slider fast you will notice a weird tape slowing style effect...

indeed. a long release time helps. now only the transients seem to be distorted and their decay is fine. but you know, this is kind of weird. the whole thing sounds better without the limiter than with. and i mean, i'm not talking about strong input signals. it's just peaking slighly of +4.

bests,
salat

[salatmensch]

hiya,

the oscilator waveform view doesn't seem to respond to the "shift+click" method. for example in the freqshifter-module. when i shift+click on "dry" a small value entry window appears. when i click on "main shape" that window doesn't come. the only reason why i'm missing that window is because it's really difficult to set the main shape to "100% noise" for example. i only manage to "98% noise/ 2% mess".

bests,
salat

[MeldaProduction]

Well, to me the limiter sounds just fine. Of course if your audio is constantly +10dB, then you just cannot assume it will fix it. Maybe you can send me some audio, so I can check it out?

Saturation is great, but do not overuse it!

Also keep Attack at minimum in these cases. Release shouldn't be too high as then it will constantly reduce the gain, which is usually not desired. Limiters are usually used to remove peaks and increase overall loudness, not to stabilize volume. Use compressor or auto-volume instead.

[MeldaProduction]

hiya,

i managed to make the matrix window of the vocoder inaccessable. this is how i managed to do this. first i set up the number of bands to 100 and opened and closed the window. it barely fitted the screen (have a resolution of 1680x1050). then i reduced the number of bands to 32. now i opened the window and i can barely see the ok button. most of the window has disappeared in the virtual realm above the screen. the problem now is that i cannot move the window down because i don't have access to the top of the window. removing mxxx and putting it back in doesn't affect the window position. it will stay beyond my screen size.

bests,
salat

yej, nice, will check it out

[MeldaProduction]

hiya,

the oscilator waveform view doesn't seem to respond to the "shift+click" method. for example in the freqshifter-module. when i shift+click on "dry" a small value entry window appears. when i click on "main shape" that window doesn't come. the only reason why i'm missing that window is because it's really difficult to set the main shape to "100% noise" for example. i only manage to "98% noise/ 2% mess".

bests,
salat

Just click into the editor (no keys, just click)

[MeldaProduction]

hiya,

i managed to make the matrix window of the vocoder inaccessable. this is how i managed to do this. first i set up the number of bands to 100 and opened and closed the window. it barely fitted the screen (have a resolution of 1680x1050). then i reduced the number of bands to 32. now i opened the window and i can barely see the ok button. most of the window has disappeared in the virtual realm above the screen. the problem now is that i cannot move the window down because i don't have access to the top of the window. removing mxxx and putting it back in doesn't affect the window position. it will stay beyond my screen size.

bests,
salat

yej, nice, will check it out

Ok, just search for file "VocoderBandMatrix.winstate" and delete it. It will start working again. In the next version, only 50 items will be displayed in the matrix, so it should fit all screens.