What image dimensions actually mean

Pixels are the only unit that matters here

A digital image is a grid of coloured squares. Its dimensions are how many squares across and how many down — 4032 × 3024, say. That is the complete description of how much visual information the file contains, and it is what every upload form means when it asks for dimensions.

Multiply the two and you get the pixel count, which is where megapixels come from: 4032 × 3024 is about 12.2 million pixels, so a 12-megapixel photograph. Marketing departments prefer megapixels because it is one number instead of two, but it hides the shape, which is often the thing that matters.

Aspect ratio is the shape

The aspect ratio is the relationship between width and height, reduced to its simplest form. 4032 × 3024 divides down to 4:3. A 1920 × 1080 image is 16:9. A square is 1:1.

Two images can have wildly different pixel counts and the same ratio — 1920 × 1080 and 3840 × 2160 are both 16:9, one simply has four times as many pixels. This is why a form asking for “16:9” is asking a different question from one asking for “1920 × 1080”. The first is about shape; the second is about shape and size.

When an image is put into a frame of a different ratio, something has to give. Either it gets cropped, or it gets stretched, or it gets letterboxed with bars. Cropping deliberately, in advance, is the only one of those three you control.

DPI is not about your file

This causes more confusion than anything else in the subject. DPI — dots per inch, sometimes PPI for pixels per inch — is a number stored in the file’s metadata that says how large the image should be when printed. It has no effect whatsoever on how the image appears on a screen, and it does not change the file size.

A 3000 × 2000 image at 300 DPI and the same image at 72 DPI are byte-for-byte identical in their picture data. The only difference is a note saying “print me at 10 × 6.7 inches” versus “print me at 41.7 × 27.8 inches”.

So if a form asks for “300 DPI”, what it almost always means is “big enough to print well”. Work out the print size in inches, multiply by 300, and that is the pixel dimension you actually need. A photograph for a 6 × 4 inch print at 300 DPI needs to be 1800 × 1200 pixels.

Screen pixels are not file pixels

Modern displays pack more physical pixels into the same space than older ones. A phone might report itself as 390 points wide while having 1170 actual pixels across — three physical pixels for every one the layout uses.

This is why an image that looks sharp on an old monitor can look soft on a recent phone, and why images intended to be displayed at 400 pixels wide are often supplied at 800 or 1200. The layout says 400; the screen has more pixels than that to fill.

How dimensions relate to file size

Loosely, file size scales with the pixel count rather than the width. Halving both dimensions quarters the pixels, and typically lands somewhere near a quarter of the file size for a photograph. Doubling both quadruples it.

This is worth internalising, because it means resolution is a powerful lever. Going from 4000 pixels wide to 2000 does not make the file half the size — it makes it roughly a quarter. That is why trading a modest amount of resolution buys back so much quality when you are up against a byte limit.

What to do when a form gives you numbers

If it names both dimensions, it wants exactly that; expect a crop unless your image already has that ratio. If it gives a minimum, anything larger is fine. If it gives a ratio, the shape is what it cares about and you should choose a sensible size yourself. And if it mentions DPI, translate it into pixels before doing anything else.