print · resolution · dpi · how-to

DPI, PPI and print size: the guide that finally makes it click

Why 300 DPI is meaningless without a physical size, what the number in your file actually does, and how to work out what resolution you really need.

The OpusImg team8 min read

Someone asks for a photo at 300 DPI. You open the file, find it says 72, and change the number to 300. The file size does not move. Nothing appears different. Have you done anything?

No. And understanding why is the whole subject.

The one idea everything else follows from

An image file contains a number of pixels. It does not have a physical size.

A 3000 × 2000 photograph is three thousand pixels wide. That is all. It is not eight inches wide, or ten, or twenty. It becomes a physical size only when something decides how many of those pixels to put in each inch.

DPI is that decision. It is not a property the image has; it is a ratio between the pixels you have and the space you want them to occupy.

physical size = pixels ÷ DPI
pixels needed = physical size × DPI

That is the entire mathematics. Every confusion in this area comes from treating DPI as an attribute of the file rather than as a conversion factor.

So the same 3000 × 2000 file is:

AtPrints at
300 DPI10 × 6.7 inches
150 DPI20 × 13.3 inches
72 DPI41.7 × 27.8 inches

Same file. Same pixels. Same bytes on disk. Three completely different physical objects, because DPI is the only thing that changed and DPI lives outside the pixels.

DPI, PPI, and why the distinction matters exactly once

Strictly:

  • PPI — pixels per inch — describes an image or a screen. Pixels are the square coloured units your file is made of.
  • DPI — dots per inch — describes a printer. Dots are ink droplets, and printers use several of them to reproduce one pixel's colour.

An inkjet printing at "1440 DPI" is laying down 1440 ink droplets per inch to reproduce an image at perhaps 300 PPI. The numbers are not comparable and the printer's figure is not something you can supply.

In everyday use, everyone says DPI for both, and when a print shop asks for 300 DPI they mean 300 PPI in your file. This article uses DPI in that everyday sense, because insisting otherwise helps nobody.

The distinction earns its keep in exactly one place: when a printer's specification quotes an enormous number, that number is about ink, not about what you should supply. Sending a 1440 PPI image achieves nothing except a very large file.

What the DPI tag in your file actually does

Most image formats can store a DPI value in their metadata. It changes no pixels. It is a note attached to the file saying "when printing this, the author suggests this many pixels per inch".

Which means:

  • Changing the DPI tag without resampling changes nothing about the image. It changes the default physical size an application proposes when you print.
  • Software may ignore the tag entirely. Browsers do. Most layout applications let you place an image at any size you like, tag notwithstanding.
  • "Save for web at 72 DPI" is cargo cult. Screens do not read the tag. A 72 DPI web image and a 300 DPI web image with identical pixel dimensions are the same file with a different note. The advice originated when 72 PPI genuinely described a Macintosh display and has outlived its meaning by decades.

The tag matters when the receiving system uses it: some print workflows, some document layout applications, and some upload forms that validate it. That is why a passport photo specification says "300 DPI" and means it — the system is going to read the tag and reject a file that says otherwise, even if the pixel dimensions are correct.

The number you actually need

The famous 300 DPI figure is not arbitrary. It is roughly the point where the human eye at normal reading distance stops resolving individual dots. Above it you are spending pixels nobody perceives.

Crucially, that is at reading distance. The requirement drops as viewing distance grows, because what your eye actually resolves is an angle, not an absolute size.

OutputViewing distanceDPI needed
Business card, book, magazine25–40 cm300
Poster held or read closely40–60 cm200–300
Poster on a wall1–2 m150
Exhibition banner2–4 m100
Trade show backdrop3–5 m72–100
Billboard20 m+10–20

Billboards genuinely are printed at around 15 DPI. From a passing car it looks perfect, because at that distance your eye cannot resolve anything finer.

The practical consequence: do not supply 300 DPI for everything. A 2 × 1 metre banner at 300 DPI is a 23,600 × 11,800 pixel file — around 800 megapixels, likely several gigabytes, and completely indistinguishable at the distance it will be viewed from. At 100 DPI the same banner is 7,900 × 3,900, which is a reasonable file that no one will ever be able to tell apart from the enormous one.

Working it out in practice

"I have this image. How big can I print it?"

Divide each pixel dimension by your target DPI.

A 4000 × 3000 photo from a 12-megapixel phone:

  • At 300 DPI: 13.3 × 10 inches. A comfortable A4-plus print.
  • At 150 DPI: 26.7 × 20 inches. A wall poster, viewed from a metre.
  • At 100 DPI: 40 × 30 inches. A large banner.

"I need this physical size. What resolution do I need?"

Multiply by DPI. An A4 page (8.27 × 11.69 inches) at 300 DPI needs 2481 × 3507 pixels. A 6 × 4 inch print at 300 DPI needs 1800 × 1200.

"My file is too big for the target."

Resize it down to the exact pixel dimensions you calculated. Downscaling is essentially free in quality terms — you are discarding information you did not need, and a good resampler produces a clean result.

"My file is too small for the target."

This is the real problem, and there are three honest options:

  1. Print smaller. The most reliable fix and the one nobody wants to hear.
  2. Accept a lower DPI, if the viewing distance permits. Check the table above before assuming you cannot.
  3. Upscale. Traditional interpolation makes it bigger without adding detail. An AI upscaler generates plausible detail, which works well for photographic texture and badly for anything containing text — the trade-offs are in does AI upscaling actually work.

What does not work is changing the DPI tag from 72 to 300 and sending it. The pixel count is unchanged, and the print shop's software will compute the same physical size it always would have.

Bleed, margins and the things that eat your pixels

If the design runs to the edge of the paper, printers require bleed — usually 3 mm on every side — because paper shifts fractionally during cutting. Without it you get a hairline of white along an edge.

So an A4 poster is not 210 × 297 mm of image. It is 216 × 303 mm, of which the outer 3 mm will be trimmed away. At 300 DPI that is 2551 × 3579 pixels rather than 2480 × 3508.

Keep anything important — text, faces, logos — at least 5 mm inside the trim line as well. Between bleed and safe margin, roughly the outer 8 mm of a design is either discarded or at risk.

Scanning: the one place DPI is a real setting

When scanning, DPI is not metadata. It genuinely controls how finely the sensor samples the original, and it is the one context where the number does real work.

ScanningUse
Text document for reading or OCR300
Photograph to reprint at the same size300–600
Photograph to enlarge600–1200
35 mm film or slides2400–4800

Film needs those extreme numbers because the original is tiny. A 35 mm frame is about 1.4 inches wide; at 2400 DPI it yields around 3400 pixels, which is enough for a decent print. At 300 DPI it yields 420 pixels, which is a thumbnail.

Scanning higher than you need costs time and disk and adds nothing. Scanning lower than you need cannot be undone — the detail was never sampled.

For document scanning specifically, 300 DPI is the practical floor for reliable text extraction; below it, character shapes start to break down and recognition accuracy falls off sharply.

Common questions, answered directly

"Is 72 DPI bad for the web?" The number is irrelevant on the web. What matters is pixel dimensions — the width the image is displayed at, doubled for high-density screens.

"Should I change 72 to 300 before printing?" Only if the receiving system reads the tag and requires it, and only with resampling if you also need more pixels. Changing the tag alone changes nothing physical.

"Why does my print look soft when the file says 300 DPI?" Almost certainly because the file was upscaled to reach 300 DPI. The tag says 300; the real detail is whatever the smaller original had. The tag describes intent, not information content.

"Is 600 DPI better than 300 for printing?" Not perceptibly, for photographs at normal distance. For line art, small text and technical drawings there is a real benefit — hard edges are where extra resolution shows.

"My phone photo is 12 MP. Is that enough?" For anything up to about A3 at 300 DPI, yes. Beyond that, drop the DPI according to viewing distance and it remains fine well past poster size.

The mental model to keep

Stop thinking of DPI as a setting on your image, and start thinking of it as the answer to a division.

You have pixels. You have a physical size. DPI is what connects them, and you can only ever fix two of the three. If the physical size is fixed and the DPI requirement is fixed, then the pixel count is determined — and if your file does not have that many pixels, no metadata edit will conjure them.

Everything else in this article is a consequence of that one sentence.

We build the browser-based image tools this guide refers to, which is also how we know where they stop being the right answer.