PDF to JPG

150 dpi for anything read on a screen, 300 if it will be printed. The default in most tools is 72, which is where the "why is it blurry" question comes from: 72 dpi is a screen measurement from the 1980s and no current display is that coarse.

A PDF page has no resolution of its own. It's a description of shapes and text, so it can be drawn at any size, and the number you choose decides how many pixels you get. An A4 page at 150 dpi is about 1240 by 1754.

What each setting is for

72 dpi is a thumbnail. Fine for a preview grid, unreadable for body text.

150 dpi is the useful default. Text is comfortably legible on a screen and the files stay reasonable.

300 dpi is print quality, and what a printer or a print shop expects. An A4 page comes out around 2480 by 3508.

600 dpi is for scanning workflows and archival reproduction. The files are large and the difference is invisible on a screen.

Doubling the dpi roughly quadruples the pixel count, so the jump from 150 to 300 is not a small one.

JPG or PNG

This matters more than people expect for pages of text.

JPG suits pages that are mostly photographs. It's much smaller and the compression artefacts hide well in continuous tone.

PNG suits pages that are mostly text, line art or diagrams. JPG puts a faint halo around every hard edge, and text is nothing but hard edges. At small sizes that halo is what makes exported text look slightly dirty.

If the page is a scanned document, it's already a photograph of text and JPG is usually fine. If the page was typeset, PNG will look noticeably cleaner at the same size.

Why the margins sometimes come out black

JPG has no transparency. A PDF page is transparent wherever nothing is drawn, so converting without filling the background first leaves those areas as whatever the encoder defaults to, and black is common.

The pages here are drawn onto white first, which is what a page is. If you see black edges from another tool, that's the cause.

When exporting images is the wrong answer

If you want the words, extract the text instead. Images of text can't be searched, selected or read by a screen reader, and running recognition on an image you exported from a text PDF is a long way round to something you already had.

If you want to send the document, send the PDF. It's smaller than a folder of page images and it stays one file.

If you want to edit a page, an image is a dead end. Nothing recovers the text once it's pixels.

Exporting to images is right when something only accepts images: a slide deck, a forum that won't take PDFs, a design tool, or a page you want to crop.

Large documents

Every page is rendered separately, so a two hundred page document takes a while and produces two hundred files. The work happens on your own machine, so the limit is memory, and a long document at 300 dpi can be a lot to hold at once.

If you only need a few pages, take them out first and export those. It's faster and it avoids the memory question entirely.

Nothing is uploaded. The document is rendered in your browser, which matters given that the things people export pages from are usually contracts, statements and reports.