What DPI actually means (and which number to pick)

Every export dialog gets around to the same question sooner or later: 72, 96, 150, 200 or 300 DPI? Pick too low and the print comes out blurry. Pick too high and the file quadruples for nothing you can see. It’s easy to get right once you know what the number controls.

The definition

DPI is dots per inch: how many pixels represent one inch of physical page. It’s what ties together the two sizes an image has, its pixel dimensions and its real-world dimensions.

pixels = inches × DPI

An A4 page is 8.27 × 11.69 inches. Rendered at different DPI:

DPI	Pixel size of an A4 page	Rough use
72	595 × 842	Thumbnails, quick previews
96	794 × 1123	Screen reading
150	1240 × 1754	Good screens, casual print
200	1654 × 2339	Quality print, small text
300	2481 × 3508	Professional print, OCR, archiving

Double the DPI and you quadruple the pixel count, because it scales in both dimensions. That’s why 300 DPI files are so much bigger than 150 DPI ones, rather than just twice the size.

Where 72 and 96 come from

The odd-looking numbers are history, not engineering. Early Macintosh screens showed 72 pixels per inch, chosen so one pixel matched one typographic point (1/72 inch). That’s also why PDF’s internal unit is 1/72 inch, and why a PDF page rendered at “scale 1” comes out at 72 DPI. Windows standardised on 96 instead. Both numbers outlived the hardware that justified them, and they survive today as the conventional “screen resolution” presets in every export dialog.

Modern screens are far denser than either. A phone display goes past 400 pixels per inch. So a 72 DPI export that was fine for screens in 2005 can look soft on current devices. For anything meant to be read on screen at full size, 96 is the floor and 150 is the safe choice.

Choosing in practice

Exporting PDF pages as images, which is what the PDF to images tool does: 150 DPI is the sensible default. Sharp on any screen, and files a fraction of the 300 DPI size. Go to 300 when the image will be printed, or when you need to zoom into fine print or dense diagrams.

Compressing a scan is the same decision in reverse. How much resolution can you take away? For a document that’ll only ever be read on screen, re-rendering at 96 DPI is usually indistinguishable, and it’s where the big savings come from. Keep 150 if the original has small print in it. The PDF compressor defaults to 96 for exactly this reason.

Scanning for OCR or archiving: feed OCR engines 300 DPI. Recognition accuracy falls off quickly below that, because the engine is reading letter shapes and low resolution blurs precisely the distinctions it needs. Do the OCR first at full resolution and compress afterwards if you need to.

Print: commercial printers ask for 300 DPI at final physical size. Home inkjets are forgiving and 200 usually looks fine. Below 150, printed text has visibly ragged edges.

The classic misconception

DPI on its own says nothing about quality. It only means something alongside a physical size. A 3000-pixel-wide image is a 3000-pixel-wide image: at 10 inches wide it’s 300 DPI, at 40 inches it’s 75 DPI. Change the DPI number in a file’s metadata without resampling and you’ve changed nothing except how big software suggests printing it.

The question actually worth asking is how many pixels this use case needs. Screen reading of an A4 page wants roughly 800–1200 pixels of width. Print wants about 2500. OCR wants everything it can get. Work back from there and the DPI setting picks itself.