Control Surface Integration (CSI) Technical Feature Discussion

[jacksoonbrowne]

Hey Geoff,

I have been playing around with an ESP-32 Arduino module specifically using a rotary module with "AiEsp32RotaryEncoder" library.

I have experimented with rotary acceleration using that lib and appears acceleration is perfect in that lib.

I am going to pare down the lib to just code for acceleration.
It maybe code that could improve CSI rotary acceleration.

Whatya think? is it worth me persuing?

Cheers,
Roy

[Geoff Waddington]

Originally Posted by jacksoonbrowne

Hey Geoff,

I have been playing around with an ESP-32 Arduino module specifically using a rotary module with "AiEsp32RotaryEncoder" library.

I have experimented with rotary acceleration using that lib and appears acceleration is perfect in that lib.

I am going to pare down the lib to just code for acceleration.
It maybe code that could improve CSI rotary acceleration.

Whatya think? is it worth me persuing?

Cheers,
Roy

Don't quite understand.

CSI just works with whatever the surface delivers.

Please clarify where you are going with this.

[Geoff Waddington]

Originally Posted by MT4U

Fixed.

On the other hand, I can confirm jakeman19's reports on the rotaries.

To be more precise, the ring's and displays respond to changes in the FX UI, but any movement in the rotaries takes the parameter to 0.

Rotaries with stepped parameters work OK.

Ooops, sorry about that.

I will fix it, and then I think you will need [ 0.0 0.0 ] for your use case.

[Geoff Waddington]

New build is up.

CSI Exp.zip

Backed out stepped values change.

If you want to set to a certain value, try [ 0.0 0.0] and let me know if that works.

[MT4U]

Originally Posted by Geoff Waddington

New build is up.

CSI Exp.zip

Backed out stepped values change.

If you want to set to a certain value, try [ 0.0 0.0] and let me know if that works.

Understood. Will test it once I get to the studio.
Thanks, Geoff!

[jakeman19]

Originally Posted by Geoff Waddington

New build is up.

CSI Exp.zip

Backed out stepped values change.

If you want to set to a certain value, try [ 0.0 0.0] and let me know if that works.

Rotaries still don't work on this build for my case with this plugin.

Jd

[Geoff Waddington]

Originally Posted by Funkybot

Geoff, still getting very hard crashes here with the latest Exp. Rolling back to a more stable build in the interim as loading almost any virtual instrument plugin on a track immediately brings down Reaper.

Sorry, missed this yesterday, does this occur when CSI tries to auto map ?

[Funkybot]

Originally Posted by jakeman19

Rotaries still don't work on this build for my case with this plugin.

Jd

What type of zones are rotaries not working with? Are you talking about the new fx.zon types? Do they work in the legacy FX zones? Do rotaries work for pans in your Track.zon? Do they only not work when mapping plugins?

I ask because I had issues with Rotaries not working on my MIDI Fighter Twister when creating plugin maps using the new method, but after I refreshed CSI and loaded those maps, they worked fine. Just wondering if you're running into that.

Plus, the additional details/answers to the above may help Geoff narrow down what's wrong.

[Geoff Waddington]

New build is up.

CSI Exp.zip

Disabled auto map.

Let's see if things settle down a bit with this change.

With the new TCPFX style mapping, and the addition of the new TCPFXBuildZone functionality, auto map is of less use than before, it may be tempting fate to expect it to handle all the crazy VST's out there

[Funkybot]

Originally Posted by Geoff Waddington

Sorry, missed this yesterday, does this occur when CSI tries to auto map ?

I believe CSI was building the auto-map as soon as the plugin was loaded on the selected track. If yes, then yes. That's where it was crashing.

It seemed to occur only with plugins with large parameter counts, which makes me think it's still an overrun condition (whatever that means) like you described earlier.

I saw that you just removed the auto-map functionality, but it's incredibly great when it works and really simplifies CSI so I hope this is just temporary. I'd honestly use it way more than the SelectedTrackTCPFunctionality for FX because I'm lazy and that's fast and more than good enough for small-ish plugins.

[Geoff Waddington]

Originally Posted by Funkybot

I believe CSI was building the auto-map as soon as the plugin was loaded on the selected track. If yes, then yes. That's where it was crashing.

It seemed to occur only with plugins with large parameter counts, which makes me think it's still an overrun condition (whatever that means) like you described earlier.

I saw that you just removed the auto-map functionality, but it's incredibly great when it works and really simplifies CSI so I hope this is just temporary. I'd honestly use it way more than the SelectedTrackTCPFunctionality for FX because I'm lazy and that's fast and more than good enough for small-ish plugins.

Ok, first let's make sure everything is back to normal with the new build.

If so, maybe restrict the auto map to 8 parameters.

Seems to me the trouble started when we stopped using the "generate the first 8" approach.

[Funkybot]

Originally Posted by Geoff Waddington

Ok, first let's make sure everything is back to normal with the new build.

If so, maybe restrict the auto map to 8 parameters.

Seems to me the trouble started when we stopped using the "generate the first 8" approach.

Yep, totally understand.

You're probably already doing this, but it seems like CSI should count up the available widgets for assignment based on the SelectedTrackTCPFXTemplate, so...

Zone "SelectedTrackTCPFXTemplate"

	FXRotariesATemplate 8
	FXRotariesBTemplate 8
	FXRotariesAShiftTemplate 8
	FXRotariesBShiftTemplate 8
	FXRotariesAOptionTemplate 8
	FXRotariesBOptionTemplate 8
ZoneEnd

...48 in the above example, then just say, "ok, I'll ignore any FXParam I encounter >47" (since the param #'s start at 0).

But like I said, you're probably already doing that and I also respect that it's easier to describe than to code.

[jakeman19]

Originally Posted by Funkybot

What type of zones are rotaries not working with? Are you talking about the new fx.zon types? Do they work in the legacy FX zones? Do rotaries work for pans in your Track.zon? Do they only not work when mapping plugins?

I ask because I had issues with Rotaries not working on my MIDI Fighter Twister when creating plugin maps using the new method, but after I refreshed CSI and loaded those maps, they worked fine. Just wondering if you're running into that.

Plus, the additional details/answers to the above may help Geoff narrow down what's wrong.

Funkybot,

Legacy FX zones. Pans they work fine in track.zon. Just when I load up my VST3: CHANNEV (AnalogObsession) plugin the rotaries don't work properly. As per MT4U's comment, if you adjust via GUI on the FX and then try to use the rotary on the control surface(MCU in my case) it just jumps to lowest counter clockwise position.

Jd

[Funkybot]

Originally Posted by jakeman19

Funkybot,

Legacy FX zones. Pans they work fine in track.zon. Just when I load up my VST3: CHANNEV (AnalogObsession) plugin the rotaries don't work properly. As per MT4U's comment, if you adjust via GUI on the FX and then try to use the rotary on the control surface(MCU in my case) it just jumps to lowest counter clockwise position.

Jd

Thanks!

I'm at work and haven't had the chance to check out any builds since yesterday. Hopefully the latest Exp restores the functionality.

[jakeman19]

Originally Posted by Funkybot

Thanks!

I'm at work and haven't had the chance to check out any builds since yesterday. Hopefully the latest Exp restores the functionality.

FYI, latest build didn't work.

Jd

[Geoff Waddington]

Originally Posted by Funkybot

Yep, totally understand.

You're probably already doing this, but it seems like CSI should count up the available widgets for assignment based on the SelectedTrackTCPFXTemplate, so...

Zone "SelectedTrackTCPFXTemplate"

	FXRotariesATemplate 8
	FXRotariesBTemplate 8
	FXRotariesAShiftTemplate 8
	FXRotariesBShiftTemplate 8
	FXRotariesAOptionTemplate 8
	FXRotariesBOptionTemplate 8
ZoneEnd

...48 in the above example, then just say, "ok, I'll ignore any FXParam I encounter >47" (since the param #'s start at 0).

But like I said, you're probably already doing that and I also respect that it's easier to describe than to code.

Yup, already doing that, and tested it out.

The auto map will read the param names in very quick succession, remember the troubles we had with the clunky old auto step generator ?

Wondering it if might be a similar issue ?

[Geoff Waddington]

Originally Posted by jakeman19

Funkybot,

Legacy FX zones. Pans they work fine in track.zon. Just when I load up my VST3: CHANNEV (AnalogObsession) plugin the rotaries don't work properly. As per MT4U's comment, if you adjust via GUI on the FX and then try to use the rotary on the control surface(MCU in my case) it just jumps to lowest counter clockwise position.

Jd

Does this occur with only that plugin ?

[MixMonkey]

Originally Posted by jakeman19

Just when I load up my VST3: CHANNEV (AnalogObsession) plugin the rotaries don't work properly. As per MT4U's comment, if you adjust via GUI on the FX and then try to use the rotary on the control surface(MCU in my case) it just jumps to lowest counter clockwise position.

Do you see the same behaviour with TDR Kotelnikov? I'm not seeing this behaviour here on any plugins. I have deleted all the TCPFX stuff, though. I wanted to make sure we still had a stable point to experiment from.

[MixMonkey]

Originally Posted by jakeman19

As per MT4U's comment, if you adjust via GUI on the FX and then try to use the rotary on the control surface(MCU in my case) it just jumps to lowest counter clockwise position.

Update: This seems to be related to the presence of SubZones. Kotelnikov (no SZ) works fine. VOSEQ (two SZ) rotaries 2,3,4,5 and 7 jump to zero and stop working. Rotaries 1, 6 and 8 work normally. Go figure Investigating.

[jakeman19]

Originally Posted by Geoff Waddington

Does this occur with only that plugin ?

Others like VST3: TDR Nova (Tokyo Dawn Labs), & VST3: bx_console Focusrite SC (Plugin Alliance)

Jd

[Geoff Waddington]

Originally Posted by jakeman19

Others like VST3: TDR Nova (Tokyo Dawn Labs), & VST3: bx_console Focusrite SC (Plugin Alliance)

Jd

Ok, how about VST, as opposed to VST3 ?

[Geoff Waddington]

Originally Posted by MixMonkey

Update: This seems to be related to the presence of SubZones. Kotelnikov (no SZ) works fine. VOSEQ (two SZ) rotaries 2,3,4,5 and 7 jump to zero and stop working. Rotaries 1, 6 and 8 work normally. Go figure Investigating.

Is one of those a VST3 ?

[jakeman19]

Originally Posted by Geoff Waddington

Ok, how about VST, as opposed to VST3 ?

Seems to be a VST3 thing as I just tried VST: TDR Nova (Tokyo Dawn Labs) and this works fine.

Jd

[MixMonkey]

Originally Posted by Geoff Waddington

Is one of those a VST3 ?

No, they're both VST2. Can I suggest disabling the step sizes cache mechanism and see if this relieves the issue?

[Geoff Waddington]

Originally Posted by MixMonkey

No, they're both VST2. Can I suggest disabling the step sizes cache mechanism and see if this relieves the issue?

We re on the same page, that's why I've been asking these questions

[MixMonkey]

Originally Posted by Geoff Waddington

We re on the same page, that's why I've been asking these questions

All the same stuff (C4FX.zon) works without issue on OSC.

[Geoff Waddington]

New build is up.

CSI Exp.zip

Backed out auto generate step sizes.

Everyone please check your issues and report what you find.

[Geoff Waddington]

Originally Posted by MixMonkey

All the same stuff (C4FX.zon) works without issue on OSC.

Ah, important observation, you just gave me an idea !

[MixMonkey]

Originally Posted by Geoff Waddington

New build is up.

CSI Exp.zip

Backed out auto generate step sizes.

Everyone please check your issues and report what you find.

Seems to be back to normal here (will carry on checking)

[MixMonkey]

Pushes need to have two values specified to set a single fixed value:

RotaryPushD3 FXParam 5 [ 0.50 0.50 ]

[Geoff Waddington]

Geeez, I'm an idiot

The key phrase was "All the same stuff (C4FX.zon) works without issue on OSC".

Thanks MixMonkey !

CSI has a call from Reaper to update everything -- happens about 15-30 times a second.

When this happens CSI checks the param values for each and every plugin (along with a lot of other things) and updates the Control Surfaces.

In order to avoid message flooding, MIDI in particular, CSI checks the last sent value and only sends changes when a value has actually changed.

Because of the nature of VST plugin design, the only reliable way to determine step sizes is to turn the knobs and see what happens.

A lot of you are familiar with this method, you've used it with output monitoring on.

The automated step size calculator does exactly the same thing, at computer speed, and at very fine granularity.

CSI mutes the output whilst this is happening, otherwise, if Reaper was playing you might blow your tweeters and/or your ears as a particular parameter was swept throughout its entire range.

My stupid oversight was not considering what would be going on at the CSI outputs.

There would be an extreme flood of messages as a parameter was swept through its range, since the values would be changing they would actually be sent, OSC is very fast and could likely handle it, but MIDI would get hopelessly bogged down.

The solution is simple -- at the point where the Track is muted during the step size calculation process, CSI will also temporarily disable Control Surface feedback, avoiding the message flood.

I'll go code this up, I think this will alleviate our problems.

Stay tuned...

[Geoff Waddington]

Originally Posted by MixMonkey

Pushes need to have two values specified to set a single fixed value:

RotaryPushD3 FXParam 5 [ 0.50 0.50 ]

Yup, known issue.

[MixMonkey]

Originally Posted by Geoff Waddington

The automated step size calculator does exactly the same thing, at computer speed, and at very fine granularity.

CSI mutes the output whilst this is happening, otherwise, if Reaper was playing you might blow your tweeters and/or your ears as a particular parameter was swept throughout its entire range.

I don't want to seem too much of a Luddite, but could we please have a way of disabling all this caching/step size calculating/alias determining/automapping FX stuff in the CSI.ini?

I'm sure it'll be helpful for beginners, but it makes no sense to add the complexity to systems that already have all of these things determined manually and written to files.

[Funkybot]

Originally Posted by MixMonkey

I don't want to seem too much of a Luddite, but could we please have a way of disabling all this caching/step size calculating/alias determining/automapping FX stuff in the CSI.ini?

I'm sure it'll be helpful for beginners, but it makes no sense to add the complexity to systems that already have all of these things determined manually and written to files.

I concur. Once it's stable, I'd leave it on myself, but I think there's a good case for adding the ability to turn it off as a Page option or more global option.

[Geoff Waddington]

Originally Posted by MixMonkey

I don't want to seem too much of a Luddite, but could we please have a way of disabling all this caching/step size calculating/alias determining/automapping FX stuff in the CSI.ini?

I'm sure it'll be helpful for beginners, but it makes no sense to add the complexity to systems that already have all of these things determined manually and written to files.

Please take the following in the spirit of good natured fun

Ok, I'll make you a deal, make sure this is all working properly and, as a reward, you will get a new feature to turn it off

[Geoff Waddington]

Originally Posted by Funkybot

I concur. Once it's stable, I'd leave it on myself, but I think there's a good case for adding the ability to turn it off as a Page option or more global option.

Please see previous post

[MixMonkey]

Originally Posted by Geoff Waddington

Please take the following in the spirit of good natured fun

Ok, I'll make you a deal, make sure this is all working properly and, as a reward, you will get a new feature to turn it off

Fair enough, deal

[Geoff Waddington]

New build is up.

CSI Exp.zip

Another try at stepped value auto calculation.

You should also now be able to do this:

RotaryPushD3 FXParam 5 [ 0.50 ]

[MT4U]

Originally Posted by Geoff Waddington

New build is up.

CSI Exp.zip

Another try at stepped value auto calculation.

You should also now be able to do this:

RotaryPushD3 FXParam 5 [ 0.50 ]

Rotaries still setting a 0, but only on "some" VSTs.

All my VST3 are working perfectly.

[MT4U]

Originally Posted by Geoff Waddington

New build is up.

CSI Exp.zip

Another try at stepped value auto calculation.

You should also now be able to do this:

RotaryPushD3 FXParam 5 [ 0.50 ]

This

RotaryPushD3 FXParam 5 [ 0.50 ]

still needs repeating the value to work, but I noticed this, maybe it helps:
In the button press, the value goes to 1.00, and then back to 0.00 on button release.