Documents

Your scan is too big to upload. Fix it before you scan, not after.

Why scanned PDFs are so large, the two settings that matter more than any compressor, how to hit a specific size limit, and why uploading an ID document to a free compressor is the wrong move.

Muhammad Hasnain, founder and lead engineer of Nainnie Labs
Muhammad Hasnain6 min read

On this page

Almost every online form has a size limit, and almost every scan is over it. You scan a two-page document, the portal says maximum 2 MB, and your file is 11.

Most advice at this point is a list of compressors. The more useful fact is that the biggest reduction available to you happens before the scan exists, and it takes two settings.

Why a scan is so much bigger than a document

A PDF exported from a word processor stores text as text — characters, positions, a font reference. A scan stores a picture of the page. Those are different things by orders of magnitude.

Think about what that means per page. At 300 dpi, an A4 page is roughly 2,480 by 3,508 pixels — about 8.7 million of them. In full colour, each pixel needs 24 bits. In pure black and white, each needs 1.

That ratio is the whole article. Before any clever compression, scanning a page as bitonal rather than colour asks the file to store a twenty-fourth as much raw information. No compressor recovers that difference after the fact, because the data is already in the file and compression can only approximate its way back down.

The two settings that matter most

The US Bankruptcy Court for the Southern District of Indiana publishes a one-page handout for filers whose documents are too large, titled Tips for Reducing the Size of a PDF. Its first item is the whole of the advice above:

Scan documents using lower quality settings.

Select black and white vs grayscale or color. Lower quality to 300 dpi.

Colour mode. If the document is black text on white paper — a contract, a form, a letter, a bank statement — scan it in black and white. Keep colour only when the colour carries information: a stamp, a photo page, a highlighted clause, a coloured signature you need to prove is ink.

Resolution. 300 dpi is the standard recommendation for text and it is enough to read and to OCR. Scanning at 600 roughly quadruples the pixel count for paperwork nobody will enlarge.

Those two together routinely take a file from double-digit megabytes to under one, with no compressor involved and nothing visibly lost.

If the PDF already exists

Sometimes you cannot rescan — someone sent it to you, or the original is gone. Then compression is the tool, and it is worth knowing what it actually does to a scan: it discards image detail. Not formatting, not pages. Detail.

Which means the order of what degrades is predictable:

  1. Body text survives most compression comfortably — it is high-contrast and large.
  2. Small print goes soft well before body text does.
  3. Signatures, stamps and faint handwriting are the first casualties, because they are low-contrast and thin.

So the check after compressing is not "does it look fine" at page-fit zoom. It is: open it, zoom to 100%, and look at the signature and the smallest line of text on the page. If those are legible, you are fine. If the signature has turned into a grey smudge, you have a file that will be rejected by a human rather than by a size limit.

Hitting an exact limit

Portals do not ask for "smaller". They ask for under 2 MB, or under 500 KB, or occasionally under 100 KB.

A quality slider is a bad instrument for that. You compress, check, it is 2.3 MB, you compress again, now it is 900 KB and the signature is gone. Tools that let you name a target size and work backwards from it are worth seeking out precisely because the requirement is a number, not a feeling.

If you are stuck with a slider: go one step further than you think you need, check legibility, and step back up if there is room. Overshooting and recovering is faster than creeping down.

The part nobody mentions: you are uploading it

Here is what makes this topic different from most file-format problems.

Think about which documents actually need shrinking for an upload. Passport scans. National ID cards. Driving licences. Bank statements for a visa application. Payslips for a rental application. Tax documents. Medical letters.

That is close to a complete list of the documents you would least like a stranger to hold a copy of — and the standard advice is to upload them to a free website you found on the first page of results.

Those services are not necessarily doing anything wrong, and many delete files after an hour. But the file does leave your device, it does sit on someone's disk, and you have no way to verify what happens next. For a restaurant menu, who cares. For your passport, it is worth using something that compresses on the phone or computer itself, where the question does not arise.

When compressing is the wrong answer

When the limit is per-file and the form allows several. Split the document and upload in parts rather than crushing all of it.

When legibility is the point. If compressing far enough to fit makes the signature unreadable, you have not solved the problem, you have moved it to the person reviewing your application.

When the file is big for a different reason. A PDF that is large despite being mostly text may have an enormous embedded image on one page, or be a scan saved at 1200 dpi. Fixing the cause beats compressing the symptom.

A note on the app I build

Disclosed interest: Scandra is mine.

It is a document scanner, and the parts relevant to this post are that text recognition and all the PDF tools run on the device — so compressing a passport scan does not involve uploading a passport scan. You can compress to a target file size rather than a quality guess, which is the thing that actually matches how upload limits are written. It also merges, splits and password-protects, and exports A4, Letter, Legal or fit-to-image, all locally. There is no watermark on anything and no subscription.

The advice above works whatever you use, though. If you take one thing from it: change the colour mode before you scan, not the compression after.

Frequently asked questions

Why is my scanned PDF so large?

Because a scan is a photograph of a page, not text. A typed page saved from a word processor stores characters; a scan of the same page stores millions of pixels. That is why a five-page scan can be larger than a hundred-page document.

What dpi should I scan documents at?

300 dpi is the usual recommendation for text. A US bankruptcy court's own guidance on keeping PDFs small says to lower quality to 300 dpi and to select black and white rather than grayscale or colour. Going above 300 for ordinary paperwork adds size without adding anything you can read.

How do I compress a PDF to under 2MB?

Rescan in black and white at 300 dpi if you can, which usually does it on its own. If the PDF already exists, use a compressor that lets you set a target size rather than a vague quality slider, then open the result and check the small print and any signature are still legible.

Is it safe to use free online PDF compressors?

It depends entirely on the document. A compressor that runs in the cloud receives your file, and the documents people most often need to shrink are passports, ID cards, bank statements and tax papers. For those, use a tool that processes on your device.

Does compressing a PDF reduce quality?

For a scan, yes — compression works by throwing away image detail. Text usually survives; faint signatures, stamps and small print are what degrade first. Always open the compressed file and look at those before you submit it.

More from the blog