Have you ever seen or implemented a custom physical controller for your software via MIDI? (or some other protocol). I'm curious what super cool analog controllers exist for software that I am not aware of. Thanks much in advance!
Together with a friend I've been working on this beasty project in C++ (initially without the CS lib). We did some liveshows back then and decided that we needed some more control gear. But not just a laptop focusing launchpad, we wanted to forget there was even a laptop (or other hardware) attached to this controller, so that all our attention and focus would be on the music and the crowd, and above all: we wanted something that would instantly announce your presence on stage. So after more than a decade of work (on and off), and a lot of iterations, we have build 2 quarter circle 'cockpit style' midi controllers made from bend aluminium, spanning more than 2,5 meters combined. We've made our knob design with the idea that you would literally be able to fold your laptop and store it underneath the controller, and perform a liveset without looking at a screen.
The last bugs are still being solved, so we hope to test this in a live setting soon :) (For some pictures see: https://photos.app.goo.gl/8NHQM1Us71oJyB5B9)
https://www.lightjams.com/midiDMX.html
This makes for a much cheaper lighting controller (any old keyboard with MIDI) and lets you sync the lights to the music when part of the music is MIDI based.
Relatively new controllers such as the Linnstrument and Seaboard Rise can output MIDI MPE to control many musical parameters at the same time in expressive ways. In your thinking about unconventional uses for MIDI, MPE might give you a lot more data to work with.
I created two video camera based Theremin's to output MIDI data a long time ago, one for a PC and one for an Amiga:
http://zoom.interoscitor.com/InterOscitor/ http://zoom.interoscitor.com/PetersonEnterprises/videotherem...
I ended up using them as crude I/O devices for a Tetris-like game [2] and for simple animation tests [3]. Even ported it to WebMIDI, but hit some barriers in the amount of IO the drivers and link could handle.
At some point I must have forgotten about this, so thanks for reminding me. I might also experiment with this again. :)
Also sorry for the potato quality videos, it's the best I have at hand.
[1]: https://www.akaipro.com/akai-fire-firexus
[2]: https://www.gryphno.de/syrup/hotlink/flfire.webm (2.2M)
[3]: https://www.gryphno.de/syrup/hotlink/flfire_anim.webm (23M)
GlovePIE if I’m not mistaken lets you use a wiimote as a midi controller
TidalCycles/Supercollider lets you emit midi from haskell, there are also other libraries in other languages (rtmidi in c++? and plenty in python, also there’s the web midi api in browsers)
I once tried a project with a flight simulator joystick as an input, it was just ok though lol
pen2bow: use the apple pencil like a violin bow midi control
I also once saw a research project that used muscle-activity sensors applied externally to the throat to do /non/-acoustic pitch detection/midi control; it would tell which pitch you would be able to emit just by the posture of your vocal tract. In general I think myography seems like a cool bridge between the body and signals
I would really like to see something like a vr-based hand gesture based midi app: wave arms like symphony conductor, put some colorful shapes in 3d space to anchor movement/provide visual feedback, etc.
So you got that part figured out I suppose.
> super cool analog controllers
There is a lot of "cool" midi controllers out there. Examples:
- Modular: https://special-waves.com/ or https://jouemusic.com/
- Tactile: https://roli.com/products/blocks/lightpad-block-studio-editi...
However, I'm not sure that coolness translates to controlling non-musical apps
--
PS : If you are on windows, I find using a large touch screen to be a good tactile desktop experience (sometimes). You can even create a custom UI setup with TouchOSC.
The idea was that your password (or second factor) would be a tune, played in just the right way, with the right emphasis and velocity and even mistakes.
Bonus points, train an ML to detect nuances that define an individual's performance, as accurate as any retinal scan