DPI versus pixels, for uploads

A digital image does not have a DPI. It has a pixel count.

DPI — dots per inch — is a ratio between pixels and a physical size, and it only becomes a real number when something decides how large to print the image. The same 1200×1800 file is 300 DPI printed at 4×6 inches, 150 DPI printed at 8×12, and 600 DPI printed on a business card. Nothing about the file changed.

This matters because forms ask for DPI constantly, and understanding what they can and cannot mean by it saves a lot of wasted effort.

The metadata field, and why it does almost nothing

Image formats can store a DPI value in their metadata. JPEG has it, TIFF has it, PNG has a pixels-per-metre field.

It is advisory. It tells a printer or a layout program what physical size the author intended. It does not change the pixels, it does not change the file size, and most web upload validators never read it.

So "change the DPI to 300" in an editor, without resampling, does exactly one thing: it edits a number in the metadata. The image is pixel-for-pixel identical and the file size is unchanged to within a few bytes. If a form rejected your image and you fixed the DPI field and resubmitted, nothing of substance happened.

Some editors do something different and more dangerous: changing DPI with resampling enabled actually resizes the image, adding or removing pixels. Adding pixels is pure loss — the software invents detail that was never captured, the file gets larger, and the picture gets softer.

What a form asking for 300 DPI probably means

Three possibilities, and they want different things:

1. It wants minimum pixel dimensions and is phrasing it badly. Most common. "A 300 DPI photograph" for a 2×2 inch passport photo means 600×600 pixels. The DPI and the inches together give you the only number that exists: the pixel count. If the form states both a DPI and a physical size, multiply them and ignore the DPI from then on.

2. It is going to print your file. Then DPI is real and relevant, because something physical is about to happen. Print needs roughly 300 pixels per inch of final output for photographic quality — 240 is acceptable, 150 looks soft, 72 looks bad.

3. It is boilerplate nobody checked. Also common. The requirement was copied from a print specification into a web form, and the validator checks file size and pixel dimensions only.

The conversion, which is all you need

pixels = inches x DPI
inches = pixels / DPI

Requirement	Physical size	Pixels needed
300 DPI	2 x 2 in (passport)	600 x 600
300 DPI	35 x 45 mm (UK/EU)	413 x 531
300 DPI	A4 (8.27 x 11.69 in)	2480 x 3508
200 DPI	A4	1654 x 2339
150 DPI	A4	1240 x 1754
96 DPI	A4	794 x 1123

So "scan this A4 page at 300 DPI" means producing a 2480×3508 image — 8.7 megapixels, which is why scanned documents are so large.

Where DPI is genuinely the right unit: scanning

For scanning, DPI is the correct way to think, because there is a physical original. The scanner is deciding how many pixels to record per inch of paper, and that choice is real.

The useful figures for documents:

Scan DPI	A4 megapixels	Good for
600	35 MP	Archival, fine detail, OCR of tiny print
300	8.7 MP	The standard. Print-quality, reliable OCR
200	3.9 MP	On-screen reading, OCR of normal print
150	2.2 MP	On-screen reading, 12pt and larger
96	0.9 MP	Headings legible, body text struggling

Scanning at 600 DPI in colour when the document is black text on white paper produces a file roughly forty times larger than 150 DPI greyscale, for content that reads identically on screen.

Where DPI is the wrong unit: anything shown on a screen

Screens do not work in inches. A browser showing your photograph in a 400-pixel-wide column shows 400 pixels of it, and the metadata DPI field has no influence whatsoever.

For screen use, think only in pixels, and specifically in the largest pixel width anything will display. Double it if it might appear on a high-density display. That is the entire calculation. Resizing to that number is usually the single biggest file size saving available, and it is free.

The practical procedure for a form

1. Find the pixel dimensions in the requirement. If it gives DPI and a physical size, multiply. If it gives pixel dimensions directly, use those and ignore everything about DPI.
2. Resize to those dimensions exactly if a range is given, or to the top of the range if you have the pixels available.
3. Then hit the byte target with quality, because dimensions are now fixed and quality is the only dial left.
4. Do not touch the DPI metadata field. It changes nothing and risks triggering a resample.

That third step is where measuring matters. With dimensions locked by the form, you cannot trade pixels for bytes any more — so the only way to land just under a 200KB ceiling rather than well below it is to encode, weigh the result, and adjust.

Compress an image to 200KB, to 300KB or to 500KB bisects quality against the real byte count and ships the best result that genuinely measured under. The form specifications we have verified against a primary source are here, dimensions and format included — because a 240KB file at the wrong pixel size gets rejected just as firmly as a 400KB one.

FAQ

What does 300 DPI mean for a digital image?

On its own, nothing — a digital file has a pixel count, not a DPI. DPI is a ratio between pixels and a physical size, so it only becomes meaningful when combined with inches or millimetres. "300 DPI at 2×2 inches" means 600×600 pixels, and the pixel figure is the only one that exists in the file.

How do I change the DPI of an image?

If you mean the metadata field, most editors let you set it under image size or document properties with resampling turned off — and it will change nothing about the pixels or the file size. If you mean actually changing the resolution, you are resizing, which means choosing new pixel dimensions. Never resample upward to reach a DPI number: it invents detail, enlarges the file and softens the picture.

Does increasing DPI improve image quality?

No. Changing the metadata field does nothing at all. Resampling upward to a higher pixel count makes the file larger while making the image slightly softer, because the software is interpolating pixels that were never captured. Quality comes from how the image was captured, not from a number applied afterwards.

What DPI should I scan a document at?

200 DPI greyscale for ordinary printed text you will read on screen — comfortable for 10pt type and a quarter of the data of 300 DPI colour. Use 300 DPI for small print, handwriting, anything that will be printed, or if you intend to run OCR. 600 DPI is archival and produces files around forty times larger than 150 DPI greyscale for no on-screen benefit.