Image Resolution Explained: Pixels, DPI and Print Size

DPI is the most misunderstood number in digital imaging, largely because it does not describe how much image information you actually have. One formula clears up almost every question that follows from it.

Pixels are the only thing actually in the file

A digital image is a grid of coloured squares. An image described as 3,000 × 2,000 has 3,000 columns and 2,000 rows, and that is the complete inventory of visual information it contains. Nothing in the file says how big the picture is, because a digital image has no physical size until something displays or prints it.

This is the point where most confusion starts, so it is worth stating flatly: the same file can be a postage stamp or a billboard. Which one it becomes depends on the device rendering it, not on the file.

DPI is a ratio, not a measure of detail

PPI ("pixels per inch") describes how densely an image's pixels are packed once it has been given a physical size. It is a relationship between pixel dimensions and inches, so it is meaningless until a physical size exists. DPI ("dots per inch") strictly refers to the dots a printer lays down, which is a different thing — but in everyday workflows the two terms are used interchangeably, and most software labels the field DPI. This guide follows the common usage where it would be fussy not to.

Image files can carry a stored PPI value in their metadata, and that value is real: some software reads it to decide what physical size to assume when placing or printing the image. What it cannot do is create pixels. The amount of actual image information you have is set entirely by the pixel dimensions, and changing the stored number alone leaves those untouched.

You can see this from the other direction too: two copies of a photograph with identical pixel dimensions, one tagged 72 and one tagged 300, will print identically when printed at the same physical size. The tag decides a default, not the detail.

The only formula you need

Everything else in this guide is an application of one relationship:

An example. You have an image 3,000 pixels wide and you want to print at 300 PPI, the usual target for photographic printing. 3,000 ÷ 300 = 10 inches wide. Want it 15 inches wide instead? 3,000 ÷ 15 = 200 PPI, which is softer but often perfectly acceptable. Want 15 inches at a full 300 PPI? 15 × 300 = 4,500 pixels, which you do not have.

That is the entire mechanic. Two numbers are given, the third follows.

Working out print size

Roughly 300 PPI is the common target for photographic print quality — print shops will usually call this 300 DPI. 200 PPI is often acceptable, and below about 150 PPI softness tends to become visible in prints held in the hand.

Applying inches × 300 to common sizes gives:

These are the formula applied, not measurements — check any of them by multiplying. The practical use is in reverse: if you have a 2,000 × 1,500 image and someone wants an A3 print, you can see immediately that you are around 1,750 pixels short on the long edge, and no software setting will conjure them.

Working out screen size

Screens complicate this slightly, because modern displays pack more physical pixels into the same space than web layouts assume.

Web pages are laid out in CSS pixels, which are a unit of layout rather than a physical dot. How many physical pixels a browser uses for each CSS pixel is the device pixel ratio, and it varies by device: 1 on older displays, 2 on many phones and laptops, 3 on some phones, and fractional values such as 1.5 or 2.625 elsewhere. Browser zoom shifts it as well.

The consequence for images is direct. An image shown in a column 800 CSS pixels wide has 800 physical pixels to fill at a ratio of 1, but 1,600 at a ratio of 2 and 2,400 at a ratio of 3. Supply only 800 pixels and it will be stretched on any display above 1× and look soft.

The opposite mistake is just as common: supplying a 4,000-pixel-wide photograph for that same column means visitors download far more data than their screen can use, and the page loads more slowly for no visible gain. There is no single correct multiplier here. The right source resolution depends on which device pixel ratios you expect, whether you serve responsive images so different devices receive different files, how sharp the result needs to be, and how much file size you are willing to spend. What is worth avoiding is either extreme — one file sized for a 1× display, or one file sized for the largest screen imaginable.

Megapixels, and what the number hides

A megapixel is a million pixels. Multiply the dimensions to get it: a 4,000 × 3,000 image is 12,000,000 pixels, or 12 megapixels.

The number is a poor guide on its own, because it conceals the shape. Megapixel counts grow with the area, so doubling the linear resolution quadruples the megapixels. Going from 6 MP to 12 MP sounds like twice the detail; on each edge it is closer to 1.4 times. Since printing and display are governed by edge lengths rather than area, the dimensions are the useful figure and megapixels are the marketing one.

It also says nothing about quality. A 48 MP photo taken in poor light through a small lens can hold less usable detail than a 12 MP photo taken well. Pixel count sets an upper bound on detail; it does not deliver it.

Why upscaling does not work

If a 1,000-pixel image is too small, why not enlarge it to 3,000?

Because the software has to invent the two-thirds of the pixels that were never recorded. Traditional enlargement examines the existing pixels and interpolates plausible values between them. The result has more pixels but no more information — edges soften, fine texture disappears, and any compression artefacts get enlarged along with everything else.

Machine-learning upscalers do better by generating detail consistent with what similar images usually look like, which can be convincing for casual use. It is worth being clear about what that means: the added detail is plausible invention, not recovered fact. For a family photo, fine. For a document, an identification, or anything where the detail must be true, it is the wrong tool.

Reducing size is the opposite case and works well — discarding information is easy, and downscaling often looks cleaner than the original because it averages away noise. The asymmetry is worth internalising: down is safe, up is guesswork.

The exception: viewing distance

The 300 PPI target assumes something held at reading distance. Change the distance and the requirement changes with it, because what matters is the angle a pixel occupies in your vision.

Billboards are the extreme illustration. A large roadside poster may be printed at a fraction of the density of a magazine page and look perfectly sharp, because nobody stands a foot from it. The same logic applies at ordinary scales: a poster viewed from across a room, or a banner behind a stand, tolerates far lower density than a leaflet someone holds.

So before deciding an image is too small, ask how far away the viewer will be. It sometimes converts a problem into a non-problem.

How this affects the Workbench tools

The Image Converter keeps dimensions unless you ask it to resize. You can set a new width or height and the other side follows the original proportions. By default it will not enlarge an image, for the reason given in why upscaling does not work.

The Image to PDF tool can place images on A4 or Letter pages, or make each page match its image. PDF measures pages in points, where a point is 1/72 of an inch. On an A4 or Letter page the image is scaled to fit inside the margins, so its effective resolution is its pixel width divided by the printed width in inches. With the "match each image" option, one pixel becomes one point, so the page's nominal size works out at 72 pixels per inch.

Run the formula and the consequence appears: a 3,000 × 2,000 photograph produces a page of 3,000 × 2,000 points, which is 41.7 × 27.8 inches — around 106 × 71 cm. That is why image-sized pages are enormous on paper. Placed on A4 portrait with 10 mm margins instead, the same photo is about 190 mm (7.5 in) wide, which puts roughly 400 of its pixels into each printed inch.

On screen this is invisible, since readers fit each page to the window. When printing, choose "Fit to page" and the page scales down — which raises the effective density in the same movement. A 3,000-pixel-wide page printed across roughly 11 inches of A4 lands at about 270 pixels per inch, comfortably into good print territory. The arithmetic is the formula from the top of this guide, and it is the reason capture resolution matters when you are photographing pages to turn into a PDF.

Related