Announcing ReaLearn: Sophisticated MIDI/MCU/OSC learn for REAPER

[helgoboss]

Originally Posted by Migelius

Hi helgoboss and anyone else using this magic.

I am making a preset for my iCON Nano Platform controller. I have a few questions:

1. My display is a Mackie LCD with 8 screens. Each screen has 2 lines. But 1 line for my controller is at the bottom, the other line is at the top. If I write something long, it will transfer to the top line. Is it possible to fix this?

2. Is there a function to "center" the characters in the screen?

3. Is it possible to make long inscriptions with word transfer to the next few screens of the controller (for example, the name of the active plugin)?

4. I use the {{target.track.index}} command to specify the track number. But this command shows track number x-1. It should show 2, but it shows 1. How to fix it or maybe I use the wrong command?

1. So in your case, line 1 is at the bottom and line 2 is at the top? Then choosing the actual line should fix the issue (category "MIDI", type "Display", protocol "Mackie LCD", display x, line 1 or 2 ... not <Multiline>).

2. There's no built-in function. But it's relatively easy to program with dynamic feedback. Needs a bit basic knowledge of Lua though. Let me know if you need help with that.

3. You could choose display <All>, then long names will transfer to the next screens. But obviously this always starts from the first display, so maybe not exactly what you need.

4. Dynamic feedback to the rescue. See code below.

return {
    feedback_event = {
        value = "" .. context.prop("target.track.index") + 1
    }
}

[helgoboss]

Originally Posted by dbrozz

This was a HUGE help and allowed me to customize quite a few actions for things I needed.
Thanks so much for the advice!

A few additional questions:

- Has anyone been able to get automations working? I would like 2 faders to be dedicated to triggering volume and pan automations for the currently selected track. Is that possible?

If you map those 2 faders to "Track: Set volume" / "Track: Set pan" and choose <Selected>, that should just work. Assuming you have write/latch/touch automation mode enabled for those tracks or globally in REAPER.

Originally Posted by dbrozz

- Is there a way to get the pan setting (fader) to kind of lock to center when near? Right now with fader control only, I can't actually get center, it instead either goes 0.8% L or R.

Yes, jumping to the center if you get close can be achieved with a control transformation formula:

y = x > 0.4 && x < 0.6 ? 0.5 : x

The editor generates this image, which explains pretty nicely what this does:

Originally Posted by dbrozz

- I try and set color information for some of the pads on the controller, but it doesn't respond. It would be nice to visually differentiate the mappings Ive created.

Cheers!

You want the pads on the controller to light up in different colors? There's currently no preset for the Minilab 3 that maps ReaLearn's color information to the pad colors. What you can always do is messing with "Source min/max", which tends to change LED colors for a lot of controllers, I think the Minilab as well.

[dbrozz]

Originally Posted by helgoboss

Yes, jumping to the center if you get close can be achieved with a control transformation formula:

y = x > 0.4 && x < 0.6 ? 0.5 : x

This worked perfectly, thank you so much!

Originally Posted by helgoboss

You want the pads on the controller to light up in different colors? There's currently no preset for the Minilab 3 that maps ReaLearn's color information to the pad colors. What you can always do is messing with "Source min/max", which tends to change LED colors for a lot of controllers, I think the Minilab as well.

Yes, exactly. Minilab allows you to customize all of the pad colours but ONLY on User Scenes - which I did not use for my mapping because I had good results just using the factory Arturia Scene (Scene 1). The problem is, in this format the pads are all backlit blue, so it makes it a bit less readily apparent which pads now do what. If I could colour them, it would be easier at a glance.

I don't think the Source min/max worked, unless I was tinkering with it wrong. Do you have any specific ideas I could try?

Thanks again for the excellent information!

[audiohome]

Hi there. I have the projection in one of my pc monitors to show me what every fader does. Is there a way to have textual feedback so I know which channel is selected etc?
Or is there another way?

Thank you very much.

[audiohome]

https://imgur.com/a/sheH2kw

Also I have this problem. I press the mouse to the FX to focus it, but when I press again on Realearn it unfocuses and I can not edit it.

What do I do to change that?

Again super thank you for this amazing program.

[fazmakesmusic]

Originally Posted by helgoboss

Selector "By name" or "All by name" and then use a wildcard, e.g. "*vocals*".

Ok so going deeper with this...

I can get the first track with "vca" as prefix by using "vca*" using the Named or All name... that works great.

How can I get a bank of 8 faders to assign to each next track with the prefix "vca"?

Currently my attempts have led to 8 faders of the same track using the wildcard.

Thanks!

[7enz]

Originally Posted by audiohome

https://imgur.com/a/sheH2kw

Also I have this problem. I press the mouse to the FX to focus it, but when I press again on Realearn it unfocuses and I can not edit it.

What do I do to change that?

Again super thank you for this amazing program.

try "particular" in the fx

[soulaccess]

I have exactly the same question / issue: I use ReaLearn exclusively to target focused FX’s, and in this scenario the ReaLearn GUI is basically read-only (see audiohome‘s screencapture) - or in the worst case you click on a dial in ReaLearn and it changes the value before it disappears from view.

A typical example: when I want to finetune „step size“ for an focused FX - it is impossible without first changing settings in the target section, then changing the step size, then putting the values in the target section back to where they where, and then testing the new values to find out they are not perfect, so you start all over again, this can easily become a source of great frustration.

So for now my workaround is to edit the .json files directly.

Probably I‘m doing something obviously stupid, I would appreciate if someone could shed some light on this issue.

Originally Posted by audiohome

https://imgur.com/a/sheH2kw

Also I have this problem. I press the mouse to the FX to focus it, but when I press again on Realearn it unfocuses and I can not edit it.

What do I do to change that?

Again super thank you for this amazing program.

[helgoboss]

Originally Posted by soulaccess

I have exactly the same question / issue: I use ReaLearn exclusively to target focused FX’s, and in this scenario the ReaLearn GUI is basically read-only (see audiohome‘s screencapture) - or in the worst case you click on a dial in ReaLearn and it changes the value before it disappears from view.

A typical example: when I want to finetune „step size“ for an focused FX - it is impossible without first changing settings in the target section, then changing the step size, then putting the values in the target section back to where they where, and then testing the new values to find out they are not perfect, so you start all over again, this can easily become a source of great frustration.

So for now my workaround is to edit the .json files directly.

Probably I‘m doing something obviously stupid, I would appreciate if someone could shed some light on this issue.

You aren't doing anything stupid. What you and audiohome are experiencing, is a known issue.

It's a conceptual issue, not a technical one: You told ReaLearn to target whatever FX is currently focused. As soon as you focus ReaLearn again, the other plug-in is - as a matter of fact - not focused anymore, so it's not the targeted FX anymore. Technically, ReaLearn does the correct thing here. But I'm aware that this is in practice very annoying

The good thing about this being a conceptual issue is that everyone could actually help with it. I would like to fix this but I'm so busy with Playtime 2 development that I'm not very eager at the moment to figure out a correct and elegant solution to this issue.

So my suggestion: Whoever has this issue, try to come up with a conceptual solution and discuss it here. Ideas, please

BTW, just not considering "ReaLearn" as focusable in general would not be a proper solution! Maybe something like a "Capture currently focused plug-in while mapping panel open" kind of thing? No idea. Back to Playtime.

[helgoboss]

Originally Posted by fazmakesmusic

Ok so going deeper with this...

I can get the first track with "vca" as prefix by using "vca*" using the Named or All name... that works great.

How can I get a bank of 8 faders to assign to each next track with the prefix "vca"?

Currently my attempts have led to 8 faders of the same track using the wildcard.

Thanks!

Not possible I think. That would need some selector like "n-th by name", not "(First) by name" or "All by name".

[helgoboss]

Originally Posted by audiohome

Hi there. I have the projection in one of my pc monitors to show me what every fader does. Is there a way to have textual feedback so I know which channel is selected etc?
Or is there another way?

Thank you very much.

Not yet. Projection will be improved at some point but after Playtime 2 release.

[helgoboss]

Originally Posted by dbrozz

I don't think the Source min/max worked, unless I was tinkering with it wrong. Do you have any specific ideas I could try?

Thanks again for the excellent information!

Ah, I think the LEDs of the pads on MiniLab 2 were sys-ex controlled. Maybe it's the same for the MiniLab 3? You can use ReaLearn to send different sys-ex. "MIDI script" (Lua) is the best source for this.

[soulaccess]

Originally Posted by helgoboss

So my suggestion: Whoever has this issue, try to come up with a conceptual solution and discuss it here. Ideas, please

Thank you for your explanation; I thought it might be something along these lines.

Since you mentioned that simply ignoring the focusing of ReaLearn is not an option, I propose the following solution (which I hope doesn't sound utterly nonsensical to you):

ReaLearn should store the last focused FX (the one which was focused before you clicked - and thus focused - ReaLearn) and use that information based on the context. I identified two situations where ReaLearn should disregard being focused and align with the stored last focused FX:

1) In the main window: if "Auto-load" is set to "Based on instance FX", the "Main preset" dropdown should use the last focused FX.

2) In the mapping window: If the "FX" dropdown in the target section is set to <Focused>, the glue section should align with the stored last focused FX.

You could potentially identify other contexts, but for me, this solution would address the situation, seamlessly integrate, and feel much smoother at the same time. This would also avoid introducing an additional switch or checkbox to "change mode" (which I think should be the very last resort).

I hope I was able to present my idea clearly. Don't hesitate to discuss this further. Thank you for listening.

[Puck]

Just offering a workable solution. I find this more elegant anyhow:

Use <instance> instead of <focused>

Map a set of buttons (I actually use the F keys on my keyboard) to set a particular FX slot to the instance fx of a separate Realearn instance. I do that with a mapping that looks like this:

{
    kind = "Mapping",
    version = "2.16.0-pre.1",
    value = {
        id = "y-N0irf5bIB72xswJ1lUX",
        name = "1",
        feedback_enabled = false,
        source = {
            kind = "Key",
            keystroke = {
                modifiers = 1,
                key = 112,
            },
        },
        glue = {
            step_size_interval = {0.01, 0.05},
            step_factor_interval = {1, 5},
        },
        target = {
            kind = "Fx",
            fx = {
                address = "ByIndex",
                chain = {
                    address = "Track",
                    track = {
                        address = "Selected",
                    },
                },
                index = 0,
            },
            action = "SetAsInstanceFx",
            instance_tags = {
                "fxcontrol",
            },
        },
    },
}

The separate realearn instance where I program the plugins has a tag of "fxcontrol"

You do that by hitting "instance data" on the bottom right of Realearn and adding whatever tag you would like.

In addition to this I also have actions that I've mapped to my F keys that float the right plugin.

But this way, clicking on realearn doesn't "unfocus" your instance FX resulting in this somewhat annoying behavior.

[soulaccess]

Originally Posted by Puck

Just offering a workable solution. I find this more elegant anyhow:

Interesting solution, thank you Puck. Probably rock solid. If I understand correctly, do you always have to press a key before you can control an FX? This would be a let down for me, as it adds an additional step when going from one FX to another.

The built-in ReaLearn functionality works actually very well for switching from one FX to another (without any further help by the user). It is only a problem when you try to edit or fine-tune those settings.

[Puck]

Originally Posted by soulaccess

Interesting solution, thank you Puck. Probably rock solid. If I understand correctly, do you always have to press a key before you can control an FX? This would be a let down for me, as it adds an additional step when going from one FX to another.

The built-in ReaLearn functionality works actually very well for switching from one FX to another (without any further help by the user). It is only a problem when you try to edit or fine-tune those settings.

Yes you’d always have to press a key. I personally like it that way as I have scripts that close all other plugins and float the specific one I’m after in the center of the screen. But I can see how you wouldn’t want to have to do that and prefer the focused method. The only caveat being that making the presets has an annoying quirk.

If you are potentially mapping a lot of plugins, maybe it’s worth it to look into automating the task with LUA. But that can be a can of worms

[fazmakesmusic]

Originally Posted by helgoboss

Not possible I think. That would need some selector like "n-th by name", not "(First) by name" or "All by name".

No problem! Thanks for your reply!

Likely my need is very “edge-case”.

Wrote a script to show by name match here - more than ok for my needs.

Thanks!

[dbrozz]

Originally Posted by helgoboss

Ah, I think the LEDs of the pads on MiniLab 2 were sys-ex controlled. Maybe it's the same for the MiniLab 3? You can use ReaLearn to send different sys-ex. "MIDI script" (Lua) is the best source for this.

Thank you, sir.
How would a layman go about doing something like this?
I only plan on changing the 8 pads of Bank 2 on the main Arturia preset/scene.

[MaXyM]

Originally Posted by helgoboss

Hey, anyone of the people posting here recently and haven't got a reply yet still awaiting a reply? Let me know and I will reply, otherwise I will assume you found out by yourself or it's not relevant anymore.

Hello.

Please look at:https://forum.cockos.com/showthread....rp#post2703830

I found the original post/image I was referring too:
https://forum.cockos.com/showpost.ph...3&postcount=10


This feature is not crucial for me anymore, since I found another way to achieve that. However, I would like to satisfy my curiosity and learn more about ReaLearn. Obviously, I'm missing some bit of information which prevents me from succeeding.

Thank you in advance.

[DaniloVillanova]

Hi, thanks for the incredible script Helgoboss! This is by far the most useful REAPER extension ever IMO.

I've used TouchOSC to create a controller surface for my currently selected FX. I've successfully mapped the controller preset to my FX mapping presets and I'm struggling to get feedback to the text/value "displays" that I made on the TouchOSC template.

Each encoder/fader has two OSC "displays" next to it, each with its own address. One for the name and one for the numeric value of the currently selected FX parameter.

How can make the two displays show feedback from their associated encoders/faders?

I hope I'm explaining myself clearly.

[foxAsteria]

Originally Posted by DaniloVillanova

How can make the two displays show feedback from their associated encoders/faders?

Text expressions:
https://github.com/helgoboss/realear...ext-expression

[DaniloVillanova]

Originally Posted by foxAsteria

Text expressions:
https://github.com/helgoboss/realear...ext-expression

Thanks. When I select Text expressions in the Controller pane the text input field disappears. It only appears in the Mapping panel. Is there a way to define the string in the Controller pane or do I need to do it manually for each FX preset?

[DaniloVillanova]

This is how I set up my parameters and their respective displays in TouchOSC:

/fxparam/1/value x
/fxparam/1/value text
/fxparam/1/name text

I "learned" the controllers in the Controller Mapping panel and their addresses are recognized correctly.

Them, I paired my newly created Controller preset to my FX mappings, which ReaLearn does automatically (nice!) based on the names of the controls (I have a hardware controller that I mapped previously, and I used the same naming structure for the controls).

Expected behavior: I want the displays associated with their respective controllers to ALWAYS show the corresponding name and text value of the mapping that is active right now in the Mapping Panel.

Can I do this WITHOUT having to define the display behavior on each FX mapping? (It's hundreds of them!) I've spent hours reading the manual and days trying to set this up. Can anybody provide a solution?

Helgoboss? Anyone?

EDIT: Also, I just wanted to share this video of Tycho showing his REAPER setup with ReaLearn which I thought was very cool:

https://www.youtube.com/watch?v=5ZWe..._channel=Tycho

[helgoboss]

Originally Posted by DaniloVillanova

This is how I set up my parameters and their respective displays in TouchOSC:

/fxparam/1/value x
/fxparam/1/value text
/fxparam/1/name text

I "learned" the controllers in the Controller Mapping panel and their addresses are recognized correctly.

Them, I paired my newly created Controller preset to my FX mappings, which ReaLearn does automatically (nice!) based on the names of the controls (I have a hardware controller that I mapped previously, and I used the same naming structure for the controls).

Expected behavior: I want the displays associated with their respective controllers to ALWAYS show the corresponding name and text value of the mapping that is active right now in the Mapping Panel.

Can I do this WITHOUT having to define the display behavior on each FX mapping? (It's hundreds of them!) I've spent hours reading the manual and days trying to set this up. Can anybody provide a solution?

I have understood the following, correct me if I'm wrong:

- You have a controller preset loaded (into the controller compartment). It always stays the same.
- You load different main presets (into the main compartment), depending on which FX you want to control.
- You don't want to repeat the text feedback mappings in each main preset. Instead you want to put the text feedback mappings into the controller compartment.

Is this possible? No. For *each* desired connection between a source on your controller (knob, fader, LCD display, ...) and a target in REAPER (FX parameter value, track volume, FX parameter name, ...), you need to have exactly one mapping. There's no way around that. You *can* create 2 mappings with the same virtual target name in the controller compartment (which I assume you did) but this doesn't have the effect you desire.

The best solution in ReaLearn to avoid repetition is to use Lua (ReaLearn Script) for building your presets. That allows you to use functions and variables and what not in order to avoid repeating similar stuff again and again. I would personally always reach for Lua if things get too complex and big. It's quite easy to start with it because you can export a compartment as Lua code, so you don't need to start from scratch.

[mcconnor]

Hi to the pro's

I got myself an Arturia Beatstep because of the nice form factor. But I am struggeling to set it up even after watching the tuts (excellent by the way).

The Beatstep is ACTIVE in the Midi Settings of Reaper and is seen by the ReaLearn Plugin - I can choose it for Input and Output. But upon pressing Learn many It does not recognize any of the fadern turns or knobs pressed - nothing. I could not assign a single element. I also set the input of the channel to Midi all Channels and armed record even if I thought i had understood this was not possible but still no reaction. Am I holding it wrong?

[Perken]

Originally Posted by mcconnor

Am I holding it wrong?

In the realearn vst window, at the top, there's two checkboxes: controller compartment and main compartment.
Your midi controller must be mapped from the controller compartment (don't forget to save as a preset) and then those mappings can be re-used in the main compartment.

Did you set up the assignments in the controller compartment?

[mcconnor]

Yes, controller compartment. In the tut's it was described as beeing the right way to establish a first preset of an controller that has no preset already available and I couldnt find one for the Arturia Beatstep unfortunately. So I thought I would start right away with the correct procedure - but up to its stalled...

[Perken]

Can you share a screenshot of your realearn window?

[mcconnor]

Strange enough it works with my laptop. But not with my desktop machine. All settings were the same. It is at least a partial victory but I have no clue why it is recognized by reaper and realearn and nothing is picked up (desktop machine)...

On the laptop or seams to me sometime realearn is picky with the bindings in the main compartment - changing something may sometimes lead to the setting not responding anymore. Has to be recreated and then works... (latest version)

[DaniloVillanova]

Originally Posted by helgoboss

I have understood the following, correct me if I'm wrong:

- You have a controller preset loaded (into the controller compartment). It always stays the same.
- You load different main presets (into the main compartment), depending on which FX you want to control.
- You don't want to repeat the text feedback mappings in each main preset. Instead you want to put the text feedback mappings into the controller compartment.

Is this possible? No. For *each* desired connection between a source on your controller (knob, fader, LCD display, ...) and a target in REAPER (FX parameter value, track volume, FX parameter name, ...), you need to have exactly one mapping. There's no way around that. You *can* create 2 mappings with the same virtual target name in the controller compartment (which I assume you did) but this doesn't have the effect you desire.

The best solution in ReaLearn to avoid repetition is to use Lua (ReaLearn Script) for building your presets. That allows you to use functions and variables and what not in order to avoid repeating similar stuff again and again. I would personally always reach for Lua if things get too complex and big. It's quite easy to start with it because you can export a compartment as Lua code, so you don't need to start from scratch.

I really appreciate the detailed answer, thanks a lot. LUA is way beyond my scope unfortunately. I tried copying a mapping as LUA and pasting it in the Controller compartment and it works, but I have no idea how to make it dynamic (making it follow the target of a specific control), it just points to the mapped parameter of the particular preset.

One thing I didn't mention is that the reason I'm doing this is to use TouchOSC as "projection" display for my keyboard controller, since ReaLearn's own projection doesn't display string values. There's also a bug in it in which the parameter names don't display correctly when the text is not "wrapped" around the control in the editor. Is there any chance that the projection feature can include string values in the future? Even a single display that displays the value of the parameter that's currently being edited would be of immense help.

[Puck]

So the midi fighter twister has animations but I can’t use them because of my setup. I have two twisters linked together as one through Bome midi translator. (The reason this won’t work is those system animation messages are hard coded and I changed a lot of the midi messages so they made logical sense being one unit)

I’ve used parameter modulation to get an LED blinking, which is pretty cool

I was curious if there was a clever way to get the rainbow animation from Realearn. Ostensibly I’d like to modulate both the min and max values so it cycles through all the colors. I don’t think that’s currently possible but maybe there’s another way?

[Perken]

Originally Posted by Puck

Ostensibly I’d like to modulate both the min and max values so it cycles through all the colors.

If I recall right (haven't touched my MFT in a month), the colors activate when enable feedback for the push buttons.

[Puck]

Originally Posted by Perken

If I recall right (haven't touched my MFT in a month), the colors activate when enable feedback for the push buttons.

Yeah I have that bit down fine, thank you though!

Basically by setting the min/max you can choose what color displays (which is awesome in its own right) but if you send a certain message to the twisters you can make it do a “rainbow” animation where it cycles through the colors in a loop. That’s what I was trying to accomplish with Realearn because my setup
Doesn’t allow me to do it the way the twister wants me to.

[Perken]

mm… how about attaching one of the buttons LEDs to a dummy LFO? There's JSFx LFO for that.

[Puck]

Originally Posted by Perken

mm… how about attaching one of the buttons LEDs to a dummy LFO? There's JSFx LFO for that.

Whoops! Not sure what I was thinking!

It works just the same as the blinking. If you use parameter modulation on a realearn parameter and tie the mapping to that, then it cycles through the values (colors)

THANK YOU for pointing me in the right direction!

Since the twisters don't have displays, these little tricks will help denote special mappings

[Perken]

[jungleslam]

Hi!

I want to use realearn to receive osc messages from an external software to control the mute button in Reaper.
Realearn doesn't seem to receive any osc information, but i
i do receive incoming osc messages when listening through the port in Control/Osc/Web.
I've configured the osc input in realearn to the same port and ip used natively.
I've entered the incoming osc msg /Focus [s] "Mars" into mapping in realearn but there is no response.

What could be the issue?

Thanks
jungleslam

[mcconnor]

After having tried my luck with the Arturia Beatstep (I finally got it to work on both machines but the Encoders were too jumpy) I switched to a Midi Fighter Twister - it seems to me this is a popular und well built device and even supported by ReaLearn as Helgoboss uses it himself.

Thank you Helgoboss for this great plugin that you are developing!

Indeed it is a nice device to play with but I think I am missing some point in the configuration. I plugged in the MFT as is without changing anything to its configuration with the Tool coming with it. In Reaper after loading the Controller preset that is provided on Reapack and remapping my Main Compartment Settings to the controllers I got sometimes very strange behaviors: Sliders going down when turning the encoder up then suddenly moving in the right direction again....

I realized the Preset defines the Encoders as "Relative Type". Just to test I created a new Controller Preset. By default there the encoders are defined as "Range element". In this setting I can get my mappings to work correctly. Amazing to use with the Midi Feedback - I already feel all the possibilities along the way... Great

But I suppose Helgoboss had good reasons to create the Preset with Encoders as Relative Type and suspect probably I have to change something in the device configuration to match that. Also probably my Main compartment settings would need to be reconfigured (right now the encoders are set (glue section) to "Make relative" Type and Make absolute, no wrap.

I tried to find the infos I need to initially set this up with the correct logic in this thread but must have missed it in the 90+ pages... Could someone with experience point me to the crucial info I need to get this setup correct from the start? Thank you very much.

[loupis45]

Is it possible to use a VST parameter as a source?
For example use the CutOff from VST A as a source and Resonance in VST B as a terget

[kdarkthought]

Having a great time with ReaLearn!

Running into one snag. In the manual, it says Superiority (make instance superior) isn't necessary to handle Auto-Load because:

"Since ReaLearn 2.14.0, falling back to normal mappings when the FX loses focus in auto-load mode became much easier! One instance is enough. Your normal mappings will be memorized and reloaded once the FX loses focus. See Auto-load."

and:

"...Whenever the instance FX switches to an unlinked FX or the FX loses focus, ReaLearn falls back to the mapping list defined before activating auto-load."

I'm having trouble getting this behavior to work. I want a "default preset" loaded when an FX instance is no longer in focus. Based on what the documentation said, I made sure auto-load was switched off, sets up my default preset in "Main Preset," then turned auto-load on. I focused the FX I wanted, and auto-load worked as expected.

However, when I clicked away and the FX lost focus, the Main Preset just switched back to "<None>". What am I doing wrong here?

Thanks!