I think you can still use Scenes. You just need to plan it out, assuming you follow the same method every time.

Changing a Scene can switch on/off a Part and Change an Arp simultaneously if required.

You have 8 Scenes to play with.

If on the other hand you play it differently every time... sometimes Brass, sometimes Strings depending on your fancy, then a MIDI Foot Controller might be another option for switching Parts ON/OFF at will. You could also upgrade to a Montage... they have a dedicated row of Part buttons.

Far right row of buttons: Part button, press once.

You can now use arrows (left / right) buttons to switch parts. This is most reliable way, but careful not to press up/down buttons as these take focus off part selection.

You must have already turned off keyboard controls on all the parts you want to switch between for this approach to work.

Thanks both!

@Anthony, I do indeed want to be flexible in which parts to select while playing. Triggered by your MIDI Foot Controller suggestion, I looked for midi controllers that you can place on top of the MODX for manual control (my foot control options are exhausted with a bass pedal board, sustain pedal and expression pedal). Something like the Korg nanoKontrol 2 USB MIDI studio controller would be excellent. I'm not sure if I could actually select parts using this controller, but even if so; since it works through USB I suspect that I cannot use it in combination with my MIDI pedal board (because of the MIDI In/Out setting which has to be on either MIDI or USB). Any thoughts?

@Andrew, this is a really interesting idea that gets very close to what I'm trying to achieve. The problem is however that the part(s) that I want to keep sounding (e.g. left hand bass or chords) are also muted when I select a part - even if keyboard control was set to On for those (left hand) parts. Is there any way to exclude a Part from Part Select?

This is not great for polyphony - but an option would be to use superknob to select which part(s) are audible by setting their levels as mapped out by the "curves" using assignable knobs as sources and Part level as the output. You can use user curves to "notch out" 8 different settings for each Part. You would arrive at (as an example):

Superknob 0-15: Part A level ("on"/"off" ), Part B level ("on"/"off" ), etc ..
Superknob 16-31: ditto, making changes
Superknob 32-47: ditto, making changes
Superknob 48-63: ditto, making changes
Superknob 64-79: ditto, making changes
Superknob 80-95: ditto, making changes
Superknob 96-111: ditto, making changes
Superknob 112-127: ditto, making changes

What this sets up is a way to send a CC from an external controller (MIDI channel 1 if using multi-channel MIDI mode, or assigned MIDI IN/OUT channel if using single/hybrid) to target the setting for superknob CC. The value you would send would be maybe 0, 16, 32, 48, 64, 80, 96, or 112 depending on which combination of Part(s) you want to turn on. You can split up the range of 0-127 however you desire using user curves. I would use "step" user curves, not "linear". Set the volume of each Part you want to control to 0 and use Part volume offsets to increase the level to the desired level using the user curve for each input range.

If this is "best" or not depends. Polyphony was a stated issue. If you're already using superknob or not for other things (or if programming happens to have it tied already) may play into this. However, it does give an "easy" way to have external MIDI controllers change settings due to superknob's ability to be CC controlled (similar to scenes) and does not use scenes.

I tried to adjust the polyphony usage by using key-on delay (since it can be offset by controller assigns too) but, unfortunately, key-on delay doesn't literally delay the key-on. Polyphony is still "spent" as soon as you press a key even though the element(s) are key-on delayed. In other words, elements are "sounding"/"active" even during the key-on delay period before the tone generator actually starts any audible element output. If this was better optimized to avoid the polyphony hit - it could be used as a way to manage polyphony for sounds you set at a "0" level gaining about 8 seconds of key hold time before the particular element(s) will consume polyphony.

I found out through experience (only 13 months and counting) that the MODX, while powerful, is not limitless. It can't do everything, all of the time. Before you create a performance, you should first establish exactly what it will be used for (a particular song), and how you prefer to operate it (Arps for some parts, MIDI Controller Y/N, KB splits etc).

Only then do you have a realistic chance of building it. You need to jettison all "maybe's" from your plan, and stick to "must be's".

I still think Scene Buttons are your best option, once you have a plan.... you have 8 Scenes to "Spend", so budget accordingly.

If you want to use External MIDI Controllers to extend the "hardware ability" of the MODX, then you must buy a MIDI Controller that has the required amount of "controls" to achieve ALL the functions you need. That is "knob per function". You can't have MIDI Controller UNIT per function, nor would you want to.

MIDI Controllers add another layer of complexity and potential incompatabilities... so choose wisely.

... for external hardware - I'd probably drop my "rule" of no computer on the bandstand and use a computer with Bome or similar (for me that'd me Ctrlr and my own code - but Bome is useful as a more customer-friendly method) and pair some small USB knob/button controller (there's a few to pick from) connected to the computer. This is most flexible in terms of programming and overall cheaper than dedicated hardware solutions. Even though I like dedicated hardware solutions more - enough to "ban" a computer from my rig as of today.

Still - as warned - any turn you make into this is going to up the ante on complexity. I personally do everything with 8 scenes. Well, almost. I also automatically switch Parts when hitting a certain MIDI note using ARPs and motion sequence sometimes. This uses levels similar to the superknob suggestion (and leverages MS lanes, not superknob). That's complex too.

I agree that MODX, like any other instrument, is limited. There are plenty of walls to bump into. Also plenty of alternative routes.