This is Mods working copy with some additional fixes and features. You are welcome to use it, but the original work and tree is at https://github.com/raspberry-pi-camera/raspindi.
RaspiNDI is a simple NDI send library for Raspberry Pi. It was initially developed to allow a streaming setup for a church during the COVID-19 outbreak in 2020.
It uses the NDI library, allowing for auto-discovery of streams on the local network.
❗ This software does not work with USB cameras. It uses the native libary for the on-board camera port. ❗
This software is tested with Raspberry Pi 3B+ boards. Some users have had success running it on a Rapsberry Pi 4. The Pi Zero W does not have enough memory to run this software
Raspberry Pi 3b+ = ~600ms Raspberry Pi 4b (4GB Model Tested) = ~200ms
Against the latest relase there are three files you need:
- raspindi-${arch}
- libndioutput-${arch}.so
- libndi-${arch}.tar.gz
You will need to download the config file: wget -o /etc/raspindi.conf https://raw.githubusercontent.com/raspberry-pi-camera/raspindi/main/etc/raspindi.conf.default
Put the two sets of libraries in /usr/lib, and the raspindi binary somewhere in your PATH (eg /usr/local/bin), and run it
These intructions are for a clean installation of Raspberry Pi OS. All steps are performed on the command line.
git clone https://github.com/raspberry-pi-camera/raspindi.git && cd raspindi./easy-setup.sh/opt/raspindi/raspindi.sh
Make sure git is installed.
sudo apt update
sudo apt install git
Clone this repository and cd into it.
git clone https://github.com/raspberry-pi-camera/raspindi.git && cd raspindi
Install compilation dependencies.
sudo apt update
sudo apt install libconfig++-dev cmake libboost-program-options-dev libevent-dev libcamera-dev
Compile.
./build.sh
Install.
sudo ./install.sh
Install runtime dependencies.
sudo apt update
sudo apt install libavahi-client3 libevent-pthreads-2.1-7 libevent-core-2.1-7
Run it. (It does not require root to run.)
/opt/raspindi/raspindi.sh
Open an NDI receiver somewhere on the same network. It should detect the Raspberry Pi camera after a few seconds.
OBS Studio with the OBS-NDI plugin works well.
I added a separate "main_w_network.cpp" file that integrates a very simple API socket server with the raspindi process that allows changing some camera settings on the fly and adjust them while the stream is running.
Use system-wide libcamera-apps headers (ref: https://github.com/raspberry-pi-camera/raspindi/compare/main...nettings:raspindi:main). Enable all rpicam-vid parameters.
Updated libcamera code to fit with libcamera changes.
Fixed NeoPixel code.
Moved to using ndi 5.
Changed to using libcamera rather than MMAL directly; MMAL will not be supported in Raspberry Pi OS after the current release
Upgraded NDI library to v5.0.0
Added config parameters for various camera settings (AWB, exposure etc.)
Completely new method of acquiring the images - now calling mmal directly.
This, as well as using YUV colour, drastically improves the speed of the system,
and the smoothness of the outputted video.
Upgraded NDI library to v4.6.2
Removed OpenCV integration