JPG or PNG for a PDF?

Why the two formats behave differently

JPG is lossy. It discards detail the eye is least likely to miss, which works extremely well on photographs — gradual tonal variation, no hard edges — and very badly on sharp boundaries. A screenshot saved as JPG develops a faint halo around every letter, and that damage is permanent: re-saving cannot restore it, and each further save adds more.

PNG is lossless. Every pixel comes back exactly as it went in, and it compresses runs of identical colour extremely well, which is exactly what a screenshot or a chart is made of. On a photograph, where almost no two adjacent pixels match, it has nothing to exploit and the file is often five to ten times larger than the equivalent JPG.

So the question is not which format is better. It is which one matches the picture: continuous tone or flat colour with hard edges.

Building a PDF from images

JPG to PDF and PNG to PDF both accept either format; the names reflect the common case rather than a restriction. What matters is what your source images already are.

• Photographs from a camera or phone are already JPG. Leave them as they are — converting to PNG first cannot restore detail that JPG already discarded, and produces a much larger file for nothing.
• Screenshots should be captured as PNG and stay PNG. This is the single most common mistake in image-built PDFs: a document of screenshots saved as JPG looks blurry at exactly the moment someone zooms in to read it.
• Scans are photographs of paper, so JPG, unless the scan is pure black and white line art.

If the result is too large, compress PDF afterwards rather than degrading the images first — one controlled step is better than two uncontrolled ones. How PDF compression works explains what that step actually does.

Turning PDF pages into images

Going the other way, the same rule applies to the page you are rendering. PDF to JPG suits pages that are photographs or scans. PDF to PNG suits pages of text, diagrams or line art, where JPG artefacts land directly on the letterforms.

One thing to be clear about before converting: rendering a page to an image destroys the text. The result is pixels. It cannot be searched, selected, copied or read aloud by a screen reader, the file is usually much larger, and the change cannot be undone. That is the right trade for a thumbnail, a preview or a slide, and the wrong one for a document someone needs to read or reuse.

If what you want is the words rather than a picture of them, extract text from a PDF is the tool, and extracting text and images from a PDF covers the choice.