Note that currently, there is a small issue with regards to the H.264 decoder that is currently being worked on.
There needs to be an edit to opencv_cam for the demo. Add ros2_shared to opencv_cam/CMakeLists.txt:66.
To run a simple example, build the Dockerfile and make sure there is a file called test.mov in the directory. When executing the Dockerfile, run it with /bin/bash to get a terminal. In the terminal of the Dockerfile, run
source /opt/ros/humble/setup.bash
source install/setup.bash
ros2 launch h264_image_transport launch_decoder.py
This starts the H.264 decoder node, which will listen for H.264 messages and publish the raw decoded images.
In another terminal in the same Docker container, run
source /opt/ros/humble/setup.bash
source install/setup.bash
ros2 launch h264_image_transport launch_encoder_and_publisher.py
This starts a node that publishes a video and also launches the encoder. Note that the video may be very short, so the publisher could stop. In this case, you would simply have to restart the node.
The overall topics are /image_raw for the original image stream from the video, /in/h264 is the stream of compressed images, and /image_uncompressed is the uncompressed image stream.