Specifically you can tell matron to use a different framebuffer device as a command line argument. My original experiments with “norns on a (linux) desktop” attempted to point this at a memory mapped file and used unix fifos for the encoder/key GPIO. This is a pretty low level approach that allows the hardware emulation stuff to be in a totally separate program, so that’s kind of nice. I did get this working as an approach to running a “complete” norns stack on a laptop without any norns hardware, but I’ve since rethought the approach. I think it makes more sense to have some abstraction inside matron that lets you draw to different screen backends, which I think is more like what you want.
I’m still interested in this and will jump back into it eventually, but since taking possession of a norns I’ve been rather distracted by scripting and playing with it. For the pathologically curious my branch with this stuff is here under “hwemu”.