DPI, PPI and print size: what 300 DPI actually means

DPI and PPI are not the same thing

They are used interchangeably everywhere, including by people who know better, and in practice you can usually read one as the other. But the distinction is real and it occasionally matters.

PPI — pixels per inch — describes image pixels mapped onto a physical distance. This is the number in the arithmetic above, and it is what nearly everyone means when they say DPI.

DPI — dots per inch — describes the ink dots a printer actually lays down. An inkjet may advertise 1440 or 2880 DPI, which sounds enormously better than 300 and is not comparable at all: the printer uses many small dots of a few fixed ink colours to approximate one pixel of arbitrary colour. A 300 PPI photograph printed on a 1440 DPI printer is a perfectly ordinary configuration, not a contradiction.

The upshot: when a form says DPI it almost certainly means PPI, and a printer’s advertised DPI tells you nothing about how many pixels your file needs.

The metadata field that fools everyone

JPEG and TIFF files can store a resolution number in their metadata. This is where the confusion becomes actively harmful, because software displays that number as if it were a fact about the image.

You can tell the two apart with one question: did the pixel dimensions change? If an image was 1000 × 1000 before and is 1000 × 1000 after, nothing of substance happened, whatever the resolution field now reads.

“Save it at 72 DPI for the web”

This instruction is a fossil. It comes from early Macintosh displays, which genuinely did place roughly 72 pixels per inch, so an image at 72 PPI appeared on screen at close to its intended print size.

No current screen works that way. Phones commonly exceed 400 physical pixels per inch; desktop monitors vary by a factor of three; browsers add their own scaling on top. A browser displaying an image pays no attention to the stored resolution field whatsoever — it uses the pixel dimensions and the CSS it was given.

For the web, the stored DPI is irrelevant. What matters is pixel dimensions and file size, which is a different problem with different levers.

Why you cannot simply upscale

If 1800 pixels are needed and you have 900, the obvious move is to enlarge the image to 1800. Software will do it without complaint. It will not give you a 300 PPI print.

Enlarging resamples: it invents the new pixels by interpolating between the ones that exist. The result has the required pixel count and no additional detail, because the detail was never captured. Print it and you get a soft, slightly plastic version of the small image at the larger size. Fine detail — eyelashes, fabric texture, small text — does not reappear.

Machine-learning upscalers are better at this than classical interpolation, and they are still inventing. They produce a plausible guess at what the detail might have been, which is excellent for a wall print of a landscape and wrong for anything where accuracy matters, such as a document, a receipt or a photograph used as evidence.

The only real fix for too few pixels is a bigger original: reshoot, rescan at a higher setting, or print smaller. Printing smaller is free and usually looks better than anyone expects.

How much resolution you actually need

300 PPI is a sensible default rather than a law, and the variable it quietly assumes is viewing distance. The closer something is held, the more pixels per inch it needs.

• Held in the hand — photographs, books, brochures: 300 PPI. This is where the default comes from.
• Read at arm’s length — posters, signs: 150 PPI is usually indistinguishable.
• Seen across a room or a street — banners, billboards: 30 to 100 PPI is normal, and professionally printed billboards are often lower still.
• Line art and text — scanned documents, diagrams: 600 PPI or higher, because hard edges reveal insufficient resolution far more readily than photographs do.

If a specification demands 300 and you have 220, say so and ask rather than upscaling to hit the number. 220 real pixels per inch usually prints better than 300 with a third of them invented, and the person checking would often rather know.

Images inside a PDF

A PDF page is measured in points, and there are exactly 72 points to an inch. An A4 page is 595.28 × 841.89 points, which is 210 × 297 mm. This is fixed and has nothing to do with your image.

When you place an image on a page, its effective resolution is simply the pixels divided by the printed size:

effective PPI = image pixels ÷ (placed size in points ÷ 72)

A 1200-pixel-wide photograph spanning the full 595-point width of an A4 page is being printed across 8.27 inches, so it lands at about 145 PPI — fine for a poster, soft for a photograph. Stretching a small image to fill a page is how a PDF ends up looking worse than the picture it was made from.

This is also what Celveri’s Image to PDF tool means by the Match image size page option. Rather than fitting your image to a paper size, it makes the page the size of the image at one pixel per point — which is 72 PPI exactly. A pixel has no inherent physical size, so some figure has to be assumed, and 72 is the one that makes the page numerically identical to the image. Choose it when the PDF is for viewing on a screen; choose a real paper size when it is going to be printed.

Working it out for your own file

Three steps, no tools required beyond whatever shows you an image’s dimensions.

1. Find the pixel dimensions. Right-click and inspect the file, or open it in anything that reports them.
2. Divide each by the density you need. 3000 ÷ 300 = 10, so a 3000-pixel edge gives a good 10-inch print.
3. Compare with the print you want. If the answer is smaller, your realistic options are to print smaller or get a better original.

If you need to go the other way — you know the print size and want the pixel count — the image resizer takes exact pixel dimensions, and the table above converts the common sizes for you. For a photograph that has to fit a specific frame or aspect ratio, cropping first decides what is in the picture, and resizing afterwards decides how many pixels describe it. In that order: cropping after resizing throws away pixels you just paid for.