Drop a sample Subject, Scene and/or Style image and the AI creates a more detailed text description of each element and then recombines them. This takes away some of the hassle of creating prompts for a new AI image.
Oh – and here is your audio “deep dive” from NotebookLM.