Does raising the dpi from 72 to 300 improve image quality?

No. The dpi value is a printing instruction stored in the file header. Changing it does not add pixels or detail. It only changes how a printer or layout program interprets the physical size of the image. A 600 pixel image printed at 300 dpi is two inches wide; the same file printed at 72 dpi is about eight inches wide, with the same detail spread thinner.

How many pixels is 300 dpi at 35 by 45 mm?

413 by 531 pixels. The conversion is millimetres divided by 25.4, multiplied by 300, which gives 413.4 by 531.5 and rounds to 413 by 531. Most passport photograph specifications state the same requirement once in millimetres and once in pixels for exactly this reason.

Why does a website still reject my photo if it meets the pixel size?

Usually because the image met the dimensions only after being upscaled. Upscaling interpolates between existing samples and adds no information, so the file has the right size but visibly less detail. Other common causes are a background that is not pure white, the head occupying too small a share of the frame, or a colour profile that is not sRGB.

What 300 dpi actually means

A form asks for a photo "at 300 dpi". You open the file properties, see 72 dpi, and start looking for a setting that raises the number. There is one, and changing it will not help. This page explains what the number measures, what it does not measure, and what the form is really asking for.

The short version: dpi describes how densely to print, not how much detail a file holds. What the form actually needs is a certain number of pixels, and that count is fixed the moment the photo is taken. Being asked for 300 dpi at a given physical size is the same request as being asked for a specific pixel dimension.

The ID photo tool handles this by working backwards from the printed size you need.

Three terms that get used as if they were one

Term	What it describes	Where it lives
Pixels	How many individual samples the image contains, e.g. 600 × 600	The actual image data. Cannot be increased after the shot.
dpi (dots per inch)	How many of those samples to place in one inch when printing	A number stored in the file's metadata, usually in a few bytes.
Print size	The physical dimensions the result occupies on paper	Derived from the two above, not stored separately.

Only the first one is information. The second is an instruction about how to spend it. The third is a consequence. Once you see that, the confusion about "raising the dpi" resolves: you are not adding detail, you are telling a printer to pack the detail you already have into a smaller area.

The arithmetic, in one line

Pixels and physical size are linked by a single conversion, because there are 25.4 millimetres in an inch:

pixels = millimetres ÷ 25.4 × dpi

Worked through at 300 dpi, which is the number most forms ask for:

Printed width	Calculation	Pixels needed
25 mm	25 ÷ 25.4 × 300	295
35 mm	35 ÷ 25.4 × 300	413
45 mm	45 ÷ 25.4 × 300	531
48 mm	48 ÷ 25.4 × 300	567
50 mm	50 ÷ 25.4 × 300	591
70 mm	70 ÷ 25.4 × 300	827

This is why passport specifications look like arbitrary pixel numbers. "35 × 45 mm at 300 dpi" and "413 × 531 px" are the same requirement written two ways. The pixel pair is what actually goes on the file; the millimetre pair is what the printed document measures.

The same conversion run backwards tells you what a file is good for. A 600 × 600 pixel image printed at 300 dpi occupies 50.8 mm, exactly two inches. Print it at 150 dpi and it covers twice that, at half the density.

Why changing the dpi number does nothing

In most editors there is a box labelled resolution, showing "72" or "300", with a unit of pixels per inch. Editing that number changes one tag in the file header and leaves every pixel exactly as it was.

Two consequences follow, and they are the reason this topic causes so much wasted effort:

Setting 72 to 300 does not improve the image. The same pixels are now described as denser. Printed smaller, it will look sharper, because the same detail is packed into less space. But no new detail appeared.
Setting 300 to 72 does not damage it either. Some upload systems strip or rewrite this field, which is why the number you see on your machine and the number the site reports can differ without anything being wrong.

Where the number genuinely matters is when software decides how to place an image on a page. A word processor reading "300 dpi" will size a 600 px image as two inches wide; reading "72 dpi" it will size the same image as about eight inches wide and probably overrun the page. The pixels are identical. Only the assumption changed.

Why 300 is the number everyone asks for

It comes from commercial printing. Offset presses reproduce photographs as a grid of dots — a halftone screen — and the screen frequency is measured in lines per inch. A 150 line-per-inch screen is common for quality work. Reproducing that grid faithfully needs roughly two image samples in each direction per line, which gives 300 samples per inch.

So 300 dpi is not a quality threshold that an image either passes or fails. It is the sampling rate at which a 150 lpi halftone stops being the limiting factor. Higher than that, the press cannot show the difference. Lower, and the dot grid starts to become visible as softness.

What a form actually wants from you

When a passport office, a visa portal, or a marketplace says "300 dpi", read it as one of two things:

"Give us at least N pixels." Most often this is what is meant, and N is derivable from the physical size in the specification using the formula above.
"Do not send us a small image upscaled to look big." This is the real intent behind most such requests, and it is the failure mode that gets files rejected.

The second point is the one worth internalising. Resampling a 300 px wide image up to 413 px adds no information — the software interpolates plausible values between existing samples. The result has the right dimensions and looks soft, which is exactly the signature a reviewer is trained to spot. A 413 px image that started life at 413 px is sharp; one upscaled from 150 px is not, even though the file properties are identical.

Practical test: zoom to 100% on the face. Genuine detail holds up as texture in skin, hair and fabric weave. Upscaled detail looks smooth and slightly plastic, with edges that shimmer. If it looks plastic at 100%, the upscale will be visible in print too.

The one case where the dpi tag matters to a tool

Browsers and phones ignore dpi entirely — layout is done in CSS pixels, and a 2000 px image is 2000 px regardless of what its header claims. That is why web images are discussed in pixels and print images in dpi. The tag only becomes relevant at the boundary: when a page or a form is being prepared for printing, or when software converts between physical and pixel units using that stored assumption.

Which is also why stripping metadata before publishing a photo for sale is a good idea for a completely different reason: the same header block that carries the dpi value also carries the camera model and, very often, the GPS coordinates of where the photo was taken. The metadata remover clears that, and the EXIF guide explains what else is in there.

The conversion factor (25.4 mm per inch) and the halftone sampling relationship are standard print-industry arithmetic rather than platform policy. Published pixel figures in passport specifications are set by each issuing authority and should be checked against that authority's current guidance.