Overview

The “C++ Firmware” (consider the name a placeholder - I’ll come up with something better but that’s not a high priority until it’s a bit closer to a V1.0 release) is an ESP32 based system intended primarily for controlling interactive light, audio, and motion effects.

This kind of system isn’t ever really “finished”, although they can be abandoned. But while it’s still in “stealth mode”, it is already quite capable and feature rich - currently several hundred thousand lines of C++ code.

Current Features

No coding required

Technically this could be considered a bit of a stretch. A large part of the system functionality is driven by JSON configuration files, and those can get a bit complicated. There is also a built in DSL (Domain Specific Language)  for extra complex custom behaviors, and you have to write “code” to use it. However, this “code” is far simpler than just about anything you’d find in Arduino C++ capable of achieving similar behavior.

Also, there is no firmware to rebuild or upload with some tricky bootloader bypass process every time you want to change your project behavior. You just upload your new configuration files and go.

Wifi access REST/Websockets/Browser interface

The firmware provides

• A REST api for interactively controlling and updating the system
• A WebSockets pub-sub system
  • for realtime-ish (no polling required) monitoring of a wide (and extensible in the configuration) variety of things that change over time, including
    • system metrics (performance, memory usage, etc)
    • changing input and output values
    • project specific virtual outputs specifically for browser interface
  • for interactive browser-based control widgets
• async HTTP server for static content

There is also a Next.js/React based Web GUI which

• connects to a controller running the firmware over REST and Websockets
• provides interactive access to the controller
• 3D visualization of the controller’s “node graph”
• can be deployed directly on the controller
  • uses a Next.js static export
  • all the UI “work” happens in the browser
  • typically requires a controller with a LOT of extra flash a with a microSD interface, as the full static content is over 3MB.

Neopixel / WS2812B control

• easily control multiple (up to 8) channels. Already have a system running 1050 LEDs (7 channels with 150 LEDs each).
• positional mapping
  • every LED can be configured with a physical location o
    • partially complete OpenCV based system for automatically generating map data
  • effects can be based on
  • LED address (which is what most traditional “effects” are, like blink/fade/color chase - easy )
  • virtual canvas
    • essentially turns tree into low resolution screen
    • allows effects to use more advanced techniques like sweeps, rotations, and radial/linear fades which are effectively impossible without positional mapping

Supported I/O hardware options

Terrain Tronics

• Harlech Castle
  • with effective software based dimming support
• Caernarfon Castle

I2C

• AW9523
• MPR121
• WiiChuck
• TCA9534
• TLV493D
• VCNL4040

Audio

• I2S audio support
• virtual multi-channel mixing
• SPIFFS support for direct use of audio files stored in controller flash file system or on a microSD card

Display

• Simple I2C SSD1306 displays
• LVGL  User Interfaces
  • ILI934 TFT
  • FT6336 Capacitive Touch
  • encoder and button/key support

Planned Features

Python

• After working with CircuitPython, I’m tempted to embed a Python interpreter as an alternative to the DSL.

Private Mesh Network

This is actually already written and was the original networking mode for the firmware.

• base on painlessMesh 
• it is NOT IP networking
  • you can use painlessMesh, OR regular IP, but not both at the same time
  • Python system using painlessMesh with a python wrapper on a Raspberry Pi with a secondary USB Wifi adapter acts as an overall coordinator for for the mesh and a bridge allowing external WiFi access (through the primary Raspberry Pi WiFi connection)
• switched firmware to direct IP/REST/WebSockets as it’s much easier to get the initial setup working on a single controller
• but will probably re-enable mesh option at some point, as it offers some significant advantages for systems containing multiple controllers with some higher level interaction