Check image dimensions

The number your computer shows is not always the number everyone else sees

Asking how large an image is sounds like it should have one answer, and for most files it does. For photographs taken on a phone it frequently does not, and the gap between the two answers is where uploads get rejected and layouts break. A camera held sideways does not usually rotate the pixels it records. It stores them in the orientation the sensor saw them and writes a small tag, called EXIF Orientation, that says which way up the result should be turned. The file therefore contains one set of dimensions while every viewer displays another.

This page reports the dimensions as they are displayed. If the stored numbers differ, it says so, shows both, and names the orientation value responsible. That is deliberate: a person who checks an image against a requirement of 1080 pixels tall needs the number the receiving system will measure, not the number hidden in the header.

What the image dimensions actually tell you

Width and height in pixels are the raw answer, and they are what upload forms, ad specifications and CSS rules ask for. Aspect ratio is the shape those two numbers make once the common factor is divided out: 1920 by 1080 is 16:9, 1000 by 1000 is 1:1, 1200 by 1800 is 2:3. Ratio matters more than size when you are fitting an image into a fixed slot, because a mismatched ratio guarantees that something will be cropped or padded no matter how you scale it. That is why the two LinkedIn banner sizes are not interchangeable: a profile background is 4:1 and a Page cover is close to 6:1, so an image built for one is the wrong shape for the other however many pixels it has. The same trap catches YouTube thumbnail size, where the recommended figure is 16:9 and a square image is simply a different shape. Instagram is the instructive opposite: it accepts a whole band of ratios rather than one, and the aspect ratio reported here is what decides whether your file falls inside it — the published band is set out in Instagram post size.

Megapixels are width times height, and they describe how much work the image is rather than how it looks — a useful figure when a tab is running out of memory, since a decoded image occupies roughly four bytes per pixel regardless of how small the compressed file is. Color depth counts the bits per pixel: 24 for ordinary RGB, 32 when an alpha channel is present, 8 for grayscale. File size is the only figure here that describes the file rather than the picture, and it is the one that varies most between two images that look identical.

How large the image prints at a given DPI

Pixels do not have a physical size until you decide how densely to print them. DPI, dots per inch, is that decision: print size in inches is the pixel width divided by the DPI, and the same for height, so 3000 by 2400 pixels makes a 10 by 8 inch print at 300 DPI and a 20 by 16 inch print at 150 DPI. Drag the Print resolution slider under the answer and the size updates in inches and centimeters, ready to copy.

300 DPI is the usual target for photos, cards and anything held in the hand. Posters and signs are viewed from further away and look fine at 150 DPI or less, which is why a phone photo can make a large print that would look soft up close. The DPI number stored inside a JPEG or PNG does not enter into it: changing that tag in an editor changes no pixels, so it changes neither the sharpness nor the size this calculation gives. Only the pixel count and the density you choose do.

Read from the header, not from a decode

PNG, JPEG, GIF, BMP and WebP all record their dimensions in the first few kilobytes of the file. This tool reads only that much and answers immediately, which is why a very large image responds as quickly as a small one — nothing is decoded and no memory is spent on pixels. When the header is not one of those, the browser decodes the image instead, which covers AVIF everywhere and HEIC in Safari. The page tells you when it has taken that second path, because on that path the format details are the browser's answer rather than the file's.

Nothing is uploaded on either path. The file is opened in this tab, measured, and forgotten when you close it. If you want to confirm that for yourself rather than take it on trust, the local processing check explains what to look for in your own browser's network panel.

Reading the size and changing it are two different jobs

This tool deliberately changes nothing. There is no output file and no settings, because a measurement with options is not a measurement. When you know the numbers and want to act on them, the work belongs on a page built for it. To fit an avatar or a post slot that insists on 1:1, make the image square by cropping to the short side or padding to the long one. For a profile picture that a platform will mask into a circle, crop it to a circle and see exactly which corners disappear. When the numbers are wrong in two ways at once — too large and the wrong shape — resizing the image dimensions scales and crops in a single pass, so the picture is never resampled twice. When the same change has to be applied to a folder of images at once, the bulk image resizer does it in one pass. If the file is simply too heavy rather than the wrong shape, compressing it leaves the dimensions alone and reduces the bytes.

When you want the rest of what is in the file

Dimensions are one field among many that an image can carry. Camera model, lens, exposure, timestamps, embedded thumbnails and GPS coordinates all travel inside ordinary photographs, and the last of those is worth knowing about before sharing a picture publicly. The image metadata viewer reads those blocks and shows them field by field. Use this page when the question is how big the image is, and that one when the question is what else the image knows about you.