You'll also find that LLMs are actually quite useful in putting together scripts that will take all the photographed pages, and convert them to a proper EPUB format (using OCR, direct extraction of images, etc.), reducing the overall size and improving quality.
Visual models can even create custom CSS for you, and if there are graphs and such, LaTeX can be utilised as well.
All the tools exist separately and LLMs can write that binding glue easily.
Worth a try especially if the textbook is (regular) text heavy.