It does quite some preprocessing on the PDF pages before passing it on to tesseract.
I've found ocrmypdf to be excellent: the only issue I've had is with PDFs with differing page sizes; it seems to scale everything up to the size of the largest page, which can be a bit of a pain.