There is a number buried in almost every scan you make, and most people never see it. It decides whether the small print on a contract stays sharp, whether your phone can turn a page into searchable text, and whether the file you save today is still useful a decade from now. The number is resolution, usually written as DPI, and choosing it well is the difference between a scan that works and one that merely looks fine until the moment you need it.

What DPI Actually Measures

DPI stands for dots per inch. It describes how finely a page is sampled: how many distinct points of information are captured across each inch of paper. A scan at 300 DPI records 300 separate samples horizontally and vertically for every inch, so a standard letter-size page becomes a grid roughly 2,550 by 3,300 points wide.

Purists will point out that scanners technically produce pixels, so PPI — pixels per inch — is the more accurate term, while DPI belongs to printers laying down ink. In everyday use the two words have merged, and most scanning apps simply say DPI. What matters is the idea underneath: resolution is sampling density. The more samples you take of a curve, a serif, or a faint comma, the more faithfully the original survives the journey from paper to file.

This is why a low-resolution scan of text looks soft and stair-stepped. There simply weren't enough samples to describe the smooth edges of a letter, so the curves break into visible blocks. Add more samples and the eye stops noticing the grid at all.

Why 300 DPI Became the Quiet Standard

If you scan text at 300 DPI, the result tends to look continuous to the human eye and clean to a machine. That is not an accident. Three hundred is roughly the point where individual samples become too small for us to distinguish at normal reading distance, which is also why high-quality printing has long aimed for the same neighborhood. Below it, text starts to feel slightly degraded; above it, for ordinary printed pages, the improvements grow harder and harder to see.

The same threshold shows up in optical character recognition, the technology that converts a picture of text into actual letters you can search and copy. OCR engines are built around the assumption that characters arrive at a workable size. The open-source engine Tesseract, for example, explicitly recommends scanning at a minimum of 300 DPI, and its accuracy tends to fall off when characters get too small in the image. Feed an engine a 150 DPI scan and it has to guess from coarse, ambiguous shapes — an rn blurs into an m, a 5 slumps toward an S. Give it 300 and the same letters arrive with enough detail to be read with confidence.

So 300 DPI is less a rule someone imposed than a convergence: the resolution where human reading, print quality, and machine reading all comfortably agree. For the overwhelming majority of documents — letters, statements, forms, printed reports — it is the right answer.

When to Go Higher

There are real reasons to climb above 300, and they all come down to detail that 300 cannot quite hold.

The first is genuinely small type. Insurance riders, medical inserts, the dense footnotes of a legal agreement, and the microscopic print on warranty cards can carry text so fine that 300 DPI samples it too coarsely. At 400 or 600 DPI, those characters get more samples each, and both your eyes and an OCR engine have an easier time. If a document's value lives in its fine print, scan it like the fine print matters.

The second is archival intent. National preservation bodies have spent years writing down what "good enough to keep forever" means. In the United States, the Federal Agencies Digital Guidelines Initiative, known as FADGI, recommends scanning textual documents at around 300 to 400 PPI, and stepping higher for materials with fine detail or small features. The logic is that you scan a historical record once and then live with that file indefinitely, so it is worth capturing more than you strictly need today. If you are digitizing something irreplaceable — a deed, a family certificate, a handwritten letter — treat it as an original you may never handle again, and give it the extra resolution.

The third is anything pictorial. Photographs, stamps, fine signatures, and detailed diagrams carry information in continuous tone and texture rather than crisp letterforms, and they reward higher resolution far more than a page of body text does.

When Higher Is Just Bigger

It is tempting to assume that if 300 is good, 1,200 must be four times better. For ordinary printed text, it usually is not. Once you have enough samples to describe a letter cleanly, additional samples mostly record the paper's texture and the printer's own imperfections. You get a larger file and a longer scan, but not a more readable document.

And file size grows fast, because resolution works in two dimensions. Doubling DPI roughly quadruples the number of samples, which is why a casual jump from 300 to 600 across a stack of pages can turn a tidy archive into a sprawling one. Storage is cheap, but bloated files are slower to sync, slower to open, and slower to search. The goal is not the highest number your device allows; it is the lowest number that still preserves everything the document is worth preserving.

Resolution is also not a rescue. Scanning at 600 DPI cannot recover detail that a dim room or a shaky hand never captured. Sharp focus, even lighting, and a flat page do more for readability than any resolution setting, and they do it before resolution even enters the picture.

What This Means for Phone Scanning

Phone cameras do not have a DPI dial. They have megapixels, and the effective resolution of a "scan" depends entirely on how much of the frame the page fills. A 12-megapixel camera produces an image about 4,000 by 3,000 pixels. Aim it tightly at a letter-size page so the paper nearly fills the viewfinder, and you are capturing well over 300 effective DPI across that page. Hold the same phone far back so the document floats in a sea of desk, and you might crop down to a fraction of that — coarse, soft, and hard for OCR to read.

This is the part people miss. With a phone, framing is resolution. Filling the frame with the page, holding the camera square and steady, and letting the lens focus before you capture are what turn raw megapixels into a genuinely high-resolution document. The setting that matters most isn't in a menu; it's in how you hold the phone.

Match the Resolution to the Purpose

The honest answer to "what DPI should I scan at" is a short decision rather than a single number. For everyday documents you want to read, search, and store, 300 DPI is the dependable default. For tiny print or anything you intend to keep permanently, step up to 400 or 600. For photographs and fine visual detail, go higher still. And for almost everything else, resist the urge to climb further, because past the point of readability you are only collecting bigger files.

This is the quiet discipline behind LumenScan. It captures pages at a resolution tuned for crisp, readable documents and runs its OCR entirely on your device, so the text becomes searchable without the page ever leaving your phone — no upload, no cloud round-trip, no copy of your contract sitting on someone else's server. Good resolution and real privacy turn out to be the same instinct: capture the document faithfully, and keep it yours. If that is the kind of scanning you want, you can find LumenScan at https://lumenscan.lumenlabs.works.