Make a PDF out of something else

JPEG and PNG data goes into the document exactly as it is — no re-encoding, no loss, and no quality setting to get wrong. WebP, AVIF, GIF, BMP and HEIC have to be re-encoded once as high-quality JPEG, and only where your browser can decode them at all; HEIC generally works on iPhone and iPad and generally does not elsewhere. Anything unreadable is named in the result rather than quietly dropped.

Every mainstream HTML-to-PDF converter renders the markup in a real page in order to measure it, which means executing a stranger's file in somebody's browser. This one parses the markup as data and never inserts it into a document, so no script in it can run, no remote image or stylesheet is fetched, and javascript: links are dropped. What you get is the document typeset onto paper — headings, paragraphs, lists, tables and working links — as selectable text in a file of a few kilobytes, rather than a heavy picture of a web page.

For photographs of paperwork, matching each image keeps every page whole with no margins and no cropping, which is what somebody asking for “one PDF” usually wants. Choose A4 or US Letter instead when the result has to be printed or filed beside other paperwork; the image is then centred and scaled to fit, and a margin is worth adding if anyone will write on it.