Soundplane - source code is open source, and there was a libusb version, but it needed a few modifications/bug fixes.
Push 2 - is class compliant midi midi for buttons/encoders etc, but for the display you have to write your own driver, but their is a specification is online, and its a fairly simple protocol (its not isosynchronous)

here a demo video of Bela driving an AE Modular directly 🙂

I'll post a demo shortly of the Bela Salt doing the same thing in Eurorack ... which is also super fun.

So what are the small glitches caused by? Is it dropouts on the USB transfer, or something in the interpolation between the different capacitive sensors? Actually, do you get pre-cooked data (i.e.:: xyz) or do you get the raw sensor data and you need to make sense of it?

PS: @thetechnobear , I split this to its separate thread, hope you don't mind. Original thread here.

cool, glad you liked it ...
Bela and AE modular are a really great match, both really fun for DIY projects.

the glitches, not quite sure at the moment - I need to do analysis the data, it could be the z axis is not smoothed enough, or needs better scaling (*) - I also need to see if a different carrier set may be better (usually you run a test to see which set are most stable)

(*) the soundplane is a little 'unstable' with light touches, and it be this instability is pinging the filter/vca...as it didn't seem to happen with firmer touches. also Y is not smoothed, and I was passing this into a couple of VCAs, and a inverter... it may be rapid fluctuations there, might have been doing something to the VCA.
so, it will be interesting to compare this to Eurorack/Salt, see if the AE or patch is partly to blame 🙂

data - both cooked and raw are available, what I'm doing it using a cooked version, with the touch tracker code developed by madrona labs, as Randy has spent a lot of time developing and improving this. Its a little heavy on CPU, as its floating point... but does not have to be run at audio rate.

EDIT: interesting development...
on Salt/Eurorack, Im not getting the glitches... (except very infrequently with a one VCA and even thats unclear)

current thinking on AE/Bela:
something is accentuating a fluctuation, and pinging the vca/filter combo.. perhaps due to response rate/character of them(?!)

and/or
bela's analogs out are perhaps not as smooth as salts? a bit more raw? (perhaps salts are filtered a bit?)

probably need to log the data to check, and perhaps attach an oscilloscope too see how big the blip is...