What Does 300 DPI Actually Mean?

DPI is the most misunderstood number in image handling, and the misunderstanding is costly. People spend real time setting a file to 300 DPI, believing it has improved something, and then wonder why the print still looks soft or the form still rejects the upload.

Here is the whole of it: DPI is a printing instruction. It is one field in the file’s metadata that says “when printing this image, pack this many pixels into every inch of paper”. It changes no pixels. It changes nothing about how the image appears on a screen, because screens ignore it.

The only equation you need

pixels ÷ DPI = printed size in inches

Rearranged for the question people usually have:

printed inches × DPI = pixels required

So a photo 3000 pixels wide prints:

Same file every time. Not one pixel changed. The only thing that changed is the instruction about how tightly to pack them.

That is why “set it to 300 DPI” does nothing on its own. If the file has 3000 pixels and you tag it 300 DPI, you have declared it a 10-inch print. If you tag it 600 DPI, you have declared it a 5-inch print. The quality of the resulting print is identical at the same physical size either way, because it is the same pixels.

Why 300 became the standard

The number is not arbitrary, and it is not magic either.

Human vision at a normal reading distance of roughly 30cm resolves detail up to somewhere around 250 to 300 pixels per inch. Below that, individual dots start to become perceptible on fine detail and text edges. Above it, you are supplying detail nobody can see.

Commercial printing standardised on 300 as a comfortable figure that satisfies that threshold with margin. It is a convention based on viewing distance, which is why the sensible value changes with the viewing distance:

A billboard printed at 300 DPI would be an enormous file and would look no better from the pavement than one at 20.

What this means when a form asks for DPI

Forms that specify “3.5 × 4.5 cm at 300 DPI” are describing a pixel count in a roundabout way. They are not asking for two separate things.

Convert it: centimetres divided by 2.54 gives inches, inches multiplied by DPI gives pixels. 3.5 cm at 300 DPI is 413 pixels. The full worked version, with the common requirements pre-calculated, is in pixels, cm and DPI explained.

Then resize the actual image to that pixel count. Changing the DPI tag without resizing satisfies nothing, because almost every web validator checks pixels.

There is a narrow exception. Some print-oriented submission systems genuinely read the DPI tag and derive a physical size from it. If a system rejects a correctly sized image complaining about resolution, the tag is probably what it is reading, and setting it is then the actual fix.

Why DPI is meaningless on screens

Browsers lay out images by pixel count. A 1000-pixel-wide image occupies 1000 CSS pixels regardless of whether it is tagged 72, 300 or 1200 DPI. The field is read and ignored.

The “72 DPI for web, 300 for print” rule that circulates everywhere is a leftover. Original Macintosh displays were approximately 72 pixels per inch, so an image tagged 72 DPI displayed at roughly its printed size. No display has worked that way for decades — current screens range from around 100 to over 400 pixels per inch — and none of them consult the tag.

What does matter for screens is pixel count relative to the display. A high-density display renders a 400-CSS-pixel image using 800 or more physical pixels, so supplying only 400 produces visible softness. This is a pixel-count decision that has nothing to do with the DPI field, and it is one reason photos look soft after upload, covered in why a photo looks blurry after uploading.

Working out whether you have enough pixels

The practical question is almost always “will this print well at the size I want?”. Multiply.

A 12-megapixel phone photo is roughly 4000 × 3000 pixels. At 300 DPI that is 13.3 × 10 inches — a large print, comfortably. At 200 DPI, 20 × 15 inches, which is fine on a wall.

The common shortfall is not the camera. It is the copy. A photo that has been through a messaging app is typically reduced to around 1600 pixels on the long edge, which at 300 DPI is a 5-inch print. The original in the camera roll is fine; the version someone sent you is not. That distinction catches people preparing a print from a photo a relative shared.

If you are short of pixels, there is no honest fix. Upscaling interpolates — it invents intermediate values between existing pixels — and interpolation adds no detail. AI upscaling invents plausible detail rather than recovering real detail, which is different and occasionally matters.

Changing the DPI tag, when you actually need to

• macOS Preview: Tools → Adjust Size. Untick Resample Image, then change the Resolution field. With resampling off, only the tag changes and the pixel count stays put. With it on, the pixels are rewritten.
• Windows: Photos and Paint do not expose the field. Most third-party editors do, usually in an image size dialog alongside a resample checkbox.

That resample checkbox is the whole distinction. Unticked, you are editing metadata. Ticked, you are editing the image.

What to expect

Once the equation is in your head, DPI stops being mysterious. Decide the printed size, multiply by the resolution the viewing distance justifies, and that is the pixel count you need. If you have it, the print will be good. If you do not, no setting changes that.

The limit worth stating: nothing in a DPI field adds information to an image. The pixel count is the information. Everything else is an instruction about how to spread it across paper.