Would it be possible with a setting where the filter cutoff frequency for an instrument follows the frequency of the note being played? This makes certain types of sounds possible that are presently not possible.
also known as “Pitch Device”, already requested multiple times. Not for 2.0, so I’m moving the thread into normal suggestions section.
I think he wants this as an XRNI improvement, and there it would work on per-note basis.
this is the same problem as for Velocity Device, which only gets the latest triggered note velocity in the track; if you want polyphonic (or “polyvelocistic” 
) behaviour, you have to put two devices in two different tracks.
Brainstorming:
How could a poly device work? It should set a voices number and use a sort of first-in-first-out mechanism?
seeing notes as parameters of an enchanced XRNI structure is something we have already thought, hoping to see it implemented when we will start the XRNI2 phase
If the velocity device could catch notes aswell as velocity and there was thing called “Math Device” you could create lots of fun with it. 
The ‘poly device’ would then only be a common control device for several underlying individual devices.
For internal renoise components this could be a dynamic devices that adds the number of voices as needed just like the current instrument fx’s do. However for external plugins this would have to be a preloaded number of voices. The safest and most flexible way is to have a ‘linked’ edit function for tracks in the xrni2 structure (but also in normal pattern tracks). So when you change a parameter on one linked track, then you also change the other linked tracks. When you play notes on this instrument you have options to cycle through the different tracks you have set up in the instrument.
This link behavior could also have relative modes so you can shape individual voices slightly different for more variation.
This is by the way exactly how most older analog polyphonic synths are build (many synths inside a synth with common controls).
So for the user this would be optional if you wanna control each voice from scratch using many tracks in pattern editor for polyphony (using a cycle-track function and link devices), or if you wanna nest all this into the new xrni2 structure for better/faster overview/editing but with slightly less control of each voice.