Coincidentally I'm currently doing exactly the same thing. This is what I've learnt so far. OCR was the first problem. My photos of the book were not great; th

Coincidentally I'm currently doing exactly the same thing. This is what I've learnt so far. OCR was the first problem. My photos of the book were not great; the pages were curved in the photos and it was causing trouble for many OCR packages. Ultimately, I found that the built-in OCR in Google Photos is amazingly good, and I was able to just cut and paste the text out of the photos with barely any corrections. As for, PDF/EPUB/etc, I went for EPUB because it works better on a variety of screen sizes (it can reflow the text of course), but also because I intended to read it on my Kindle. Amazon produces free software that allows you to create books for kindle, but it will only allow you to publish directly to the Kindle store. You can't even produce a preview copy to test on your Kindle. So I abandoned that and used Calibre instead. It's OSS, and not too difficult to work with, but it works by importing Word docs or HTML files, so I had to convert my text to HTML. An EPUB file is just a zip file filled with XML, any images and some metadata, so it's easy to edit by hand. As for images: my source images are very poor quality and I've been experimenting with AI restoration and upscaling, with limited success so far. I'm proofreading the book on my Kindle at the moment, and I must say it's very satisfying.