Raspberry Ninja on NVIDIA Jetson

The Jetson Nano/NX/AGX can handle Raspberry Ninja well (1080p30 is easy compared to a Pi). These notes target JetPack 4.x images and keep the NVIDIA tegra kernel intact while updating the userland media stack under /usr/local.

Pick the right script

Most users on the provided pre-built image only need to refresh code and dependencies:

cd ~/raspberry_ninja
git pull
cd installers/nvidia_jetson
chmod +x installer.sh toolchain_update.sh setup_autostart.sh
./toolchain_update.sh

Run installer.sh when you are starting from a clean official JetPack image or repairing a badly broken system:

cd ~/raspberry_ninja/installers/nvidia_jetson
chmod +x installer.sh
sudo ./installer.sh                 # heavy build; expect several hours
# Optional legacy path if you really want an in-place distro upgrade:
# INCLUDE_DISTRO_UPGRADE=1 sudo ./installer.sh

Both flows need sudo, a stable internet connection, and plenty of free space (40 GB+ is safer). Expect to babysit long builds and rerun steps if network mirrors flake out.

Disable screen blanking and set up autostart

setup_autostart.sh can disable DPMS/console blanking, prompt for your publish.py command, and create a systemd service:

cd ~/raspberry_ninja/installers/nvidia_jetson
sudo ./setup_autostart.sh

You can rerun it to toggle screen-blanking later; only that step needs sudo.

Installing from an official NVIDIA image

The official JetPack images include the encoder/decoder bits we rely on. The installer script builds newer GStreamer and friends on top of that base. Run sections of installer.sh manually if you hit errors; newer GStreamer versions sometimes require tweaks.

Building from scratch takes hours. You can skip optional steps (SRT/FFmpeg extras) if you want a faster but less feature-complete setup.

Using the pre-built image

Latest pre-setup Jetson images (needs 16 GB uSD or larger and up-to-date firmware):

Download (updated October 2025): https://drive.google.com/file/d/1B_ywphXQ49F9we3ytcM-Zn1h7dCYOLBh/view?usp=share_link
Works on Jetson Nano 2GB A02; should also work on Nano 4GB with current firmware.
Includes GStreamer 1.26.7 with libcamera, SRT, RTMP, FFmpeg, hardware encode, and AV1 support.
Image is shrunk to ~15 GB (about 7 GB zipped); 32 GB cards are recommended.

After flashing, expand the root partition to use the full SD card (example for /dev/mmcblk0):

sudo growpart /dev/mmcblk0 1   # grows partition 1 to fill the card
sudo resize2fs /dev/mmcblk0p1  # expands the ext4 filesystem

If your device enumerates as /dev/sda, replace mmcblk0 with sda. You can also do this via GNOME Disks (“Resize…”) if you prefer a GUI.

Flash with Win32DiskImager (or balenaEtcher). Default credentials:

username: ninja
password: vdo

Chromium may not be installed on older builds.

After flashing, pull the latest code and refresh dependencies:

sudo rm -r raspberry_ninja 2>/dev/null || true
git clone https://github.com/steveseguin/raspberry_ninja
cd raspberry_ninja/installers/nvidia_jetson
sudo ./installer.sh

Missing NVIDIA GStreamer plugins?

If you are not on an official JetPack image (or something wiped the NVIDIA plugins), grab libgstnvidia.zip from this repo and restore it after running the installer:

https://github.com/steveseguin/raspberry_ninja/raw/refs/heads/main/installers/nvidia_jetson/libgstnvidia.zip
After the installer finishes, extract/copy the contents into /usr/local/lib/aarch64-linux-gnu/gstreamer-1.0/.
You can also pre-copy them to /usr/lib/aarch64-linux-gnu/gstreamer-1.0/ before running the script and hope detection picks them up.

Official NVIDIA images

Official downloads: https://developer.nvidia.com/embedded/downloads (Jetson Nano images ship with Ubuntu 18 and GStreamer 1.14; we recommend upgrading GStreamer to at least 1.16+ with the installer script).

After flashing the image and logging in:

git clone https://github.com/steveseguin/raspberry_ninja/
cd raspberry_ninja/installers/nvidia_jetson
chmod +x installer.sh
sudo ./installer.sh

Auto-start service on boot

You can adapt the Raspberry Pi service file at installers/raspberry_pi/raspininja.service if you prefer a manual service setup: https://github.com/steveseguin/raspberry_ninja/tree/main/installers/raspberry_pi#setting-up-auto-boot

If you copy it, consider changing or removing:

User=vdo
Group=vdo
Environment=XDG_RUNTIME_DIR=/run/user/1000
ExecStartPre=vcgencmd get_camera

setup_autostart.sh already handles these tweaks for Jetson users, so start there unless you need a custom service.

Details on NVIDIA’s GStreamer implementation

Docs on the encoder and pipeline options: https://docs.nvidia.com/jetson/l4t/index.html#page/Tegra%20Linux%20Driver%20Package%20Development%20Guide/accelerated_gstreamer.html#wwpID0E0A40HA

Problems and troubleshooting

“START PIPE” but nothing happens – make sure the correct video device is set (video0, video1, etc.). gst-device-monitor-1.0 lists available devices.
Other capture issues – confirm the camera supports MJPEG (or adapt the pipeline accordingly).
nvjpegdec not found – ensure the NVIDIA GStreamer plugins are present (see libgstnvidia.zip above) and that you ran the installer script successfully.

Updating firmware on Jetson Nano

Nano 2GB/4GB dev kits may need firmware updates for newer images. Updating requires Ubuntu 18 with JetPack 4 installed and can take a couple of hours. Once updated, 2GB and 4GB images are usually compatible.