How it works?

Smartpixels.app Website

The application uses the Node.js runtime - it generates a website with a control panel. Variable page elements are stored in the MySQL database which is replicated between servers. The application is based on the potential of Docker containers, it is run for production in a Kubernetes cluster. To ensure fast implementation of the code, we use the CI/CD (continuous integration and continuous deployment) with GitHub and Docker Hub services.
User page is delivered from the nearest server through Cloudflare. Production machines communicate with each other via WireGuard VPN working in a mesh configuration. Infrastructure is designed for high availability and ability to scale quickly.

Used technologies:

Node.js - is a server-side JavaScript runtime, application main programming language

Nginx - proxy server between the app and Cloudflare

Kubernetes - is an open-source platform for automating deployment and scaling application

GitHub - allows us to automate software implementation process

Cloudflare - we are using it to deliver content from the closest server to the user. It allows us to distribute load between servers

Lighting control

Colors or effects is sent to controller in a given location via Socket.io software. Each of us has different proprietary controllers, but whole thing is based on client's code written in JavaScript. For lighting we use: cards with relays, PWM brightness controllers, programmable pixel diodes, LED strips. Small Raspberry Pi computer or an ESP8266/ESP32 microcontroller is most often useful for implementation. WLED or ESP Easy software is installed on the microcontroller.

Used technologies:

Socket.io - provides two-way communication between user on website, our application and controllers in locations

Redis - is a fast database, it exchanges information about current light settings

WLED - open-source software for ESP8266/ESP32 microcontrollers - used to control programmable LEDs

ESPEasy - open-source software for ESP8266/ESP32 microcontrollers - used for PWM control of RGB LEDs and relay cards

Raspberry Pi - single board computer - usually acts as a bridge between the site and lights/LEDs

Camera video delivery

We use WebRTC and HTTP Live Streaming technologies to deliver camera footage to users. Picture from cameras are delivered to the server with graphics cards. On server, image is processed into different quality streams. OvenMediaEngine software allows us to deliver the live view with the lowest possible delay (usually 2 seconds).
As cameras, we usually use consumer surveillance IP cameras, which are calibrated to ensure the highest possible image quality.

Used technologies:

nginx-rtmp - receives incoming streams, manages authorization, and forwards them

FFmpeg - converts streams to different qualities

OvenMediaEngine - delivers WebRTC low latency streams

Nimble Streamer - delivers HTTP Live Streaming higher latency streams

Open Broadcaster Software (OBS) - allows us to apply various graphics to broadcast or mix image from few cameras

Monitoring

Our entire infrastructure and application are monitored and tested continuously. Availability monitoring is provided by Uptime Kuma, Uptime Robot, Zabbix and LibreNMS. We use Elasticsearch software to monitor application events. Hardware and performance parameters is collected by Zabbix.

Used technologies:

ELK Stack: Elasticsearch, Logstash, Kibana - application event logger

Uptime Kuma - monitors stable operation of application and quickly notifies us about problems

uptimerobot.com - backup application monitoring

LibreNMS - it is used to monitor network layer and server infrastructure

Zabbix - it is used to monitor parameters of hardware resources

Thermal printer

Idea of sharing a printer was completely spontaneous. For this task, we wanted to choose a printer that would be as maintenance-free as possible. After reviewing many offers, we chose Epson TM-T88IV printer. This printer supports the ESC / POS standard, so you can "write" to it with using many libraries available on the web.
Our proprietary script, using the DiscordPHP library, is responsible for sending messages from Discord to the printer. An appropriate message is generated and sent to RabbitMQ.
Our proprietary code is also responsible for sending messages from Tipply, which also was handled with RabbitMQ. We use RabbitMQ to queue outgoing messages to the printer. Printer is connected to computer via an RS232 port, which might not keep up with a large amount of messages - this is why we queueing messages.

Items used for this part of the project:

EPSON TM-T88IV - around 150zł

Webcamera Logitech C920 - around 350zł

Cable/Converter USB to RS232 - Unitek + adaptor DB9F/DB25M - 40zł

Graphics Card GTX 750 Ti - 420zł

Null-modem cable - DIY made

Little light source - around 25zł

Rolls (thermal paper) 80mm/80m - 5zł / pcs

SuperMicro Server (for streaming)

Intel NUC Computer (printer handler)

Software:

https://github.com/mike42/escpos-php
https://github.com/discord-php/DiscordPHP
RabbitMQ
... and our scripts

How it works?

Smartpixels.app Website

Lighting control

Camera video delivery

Monitoring

Thermal printer

Have additional questions? Want to know more?

See also our "Authors"