Hi Yvonne,
Like the others already mentioned, it is indeed better if you study a bit about Node-RED first. Afterwards if you look again at those node in the video, you will see that the author uses a series of function nodes and exec nodes.
Basically he gets a RTSP stream from his camera, and then grabs individual images from that stream. Here you can find a rather similar test flow, based on ffmpeg (instead of avconv). In that discussion you can see that the author (of your YouTube video) has also put some feedback, but on my Raspberry the performance of ffmpeg was a bit better compared to avconf. But perhaps some optimisation is somehow possible.
But good luck first with the basics of Node-RED, and then we will probably meet again here...
Bart