What DPI should I scan documents at?

200 DPI is legible for printed text and keeps files small; 300 DPI is a safe default when a reviewer may need fine detail. 600 DPI is for archival photographic reproduction and is rarely justified for a document upload.

Should I scan in colour or greyscale?

Greyscale for ordinary dark-text-on-white pages — it cuts the data to roughly a third with no meaningful loss. Colour only when a seal, stamp, ink colour or photograph matters for verification, which on official certificates it often does.

Does OCR make a scanned PDF smaller?

No, slightly larger. OCR adds a text layer on top of the existing page images. It makes the document searchable and copyable, which is valuable, but it is not a size-reduction technique.

Is a phone scan good enough for official uploads?

Usually yes, if you use a scanning app that crops to the page edges and corrects perspective, and you shoot in even indirect light. A skewed, shadowed photo of a page is rejected for legibility regardless of its resolution.

Why is my scanned PDF so much bigger than a Word export?

Because it stores a photograph of every page rather than the characters on it. A page of text is a few kilobytes as characters and several hundred kilobytes as an image of those same characters.

Can I convert a scanned PDF to text to make it smaller?

Technically yes via OCR, but you would be discarding the image of the original document. For proof-of-identity or certificate uploads that defeats the purpose, since the reviewer needs to see the document itself.

Does black-and-white scanning hurt legibility?

Not for clean printed text, where it is often the sharpest option. It handles photographs, shading and faint pencil or carbon-copy text poorly, so check the page content before choosing it.

How to Compress a Scanned PDF

Updated 12 September 2026

A scanned PDF is not really a document — it is a stack of photographs wrapped in a PDF container. That single fact explains why it behaves so differently from a Word export, and it also points to the workflow that actually makes one smaller. The leverage is almost entirely in how the scan is produced, not in what you run it through afterwards.

Resolution is the biggest lever, and it is not close

Scan resolution affects file size quadratically: doubling DPI quadruples the pixel count. A page at 600 DPI holds four times the data of the same page at 300 DPI and sixteen times that of 150 DPI. For printed text, 200 DPI is comfortably legible and 300 DPI is generous; 600 DPI is for archival reproduction of photographs and fine detail, and it is also the default on a great many scanners. Simply changing that one setting before rescanning does more than any compression tool can afford to do afterwards, because it prevents the data from existing rather than trying to throw it away later.

Choose colour mode deliberately

Colour mode is the second lever. Full colour stores three channels per pixel; greyscale stores one, cutting the data to roughly a third; black-and-white bitonal stores one bit per pixel and can be dramatically smaller still. Dark text on white paper loses nothing meaningful in greyscale. Reserve colour for documents where a coloured seal, stamp, signature ink or photograph carries verification weight — on an official certificate that is often exactly the case, so this is a judgement call rather than a rule. Bitonal is superb for clean printed text and poor for anything with shading, photographs or faint pencil.

Crop, deskew and photograph properly

If you are scanning with a phone, the surrounding desk, your hand and the shadow across the page can easily be half the frame — and every one of those pixels costs the same as a pixel of text. Use a scanning app that detects page edges, crops to them, corrects the perspective so the page is rectangular, and flattens the lighting. Shoot in indirect daylight rather than under a single overhead bulb, which casts a gradient across the page that compresses badly because JPEG has to encode the smooth tonal variation. A well-cropped, evenly lit phone scan often beats a careless flatbed scan on both size and legibility.

What OCR does and does not do

Running OCR over a scan adds a searchable, selectable text layer behind the page images. This is genuinely useful — you can search the document and copy from it — but it makes the file slightly larger, not smaller, because it adds data without removing the images. OCR only reduces size if you go further and discard the page images entirely, keeping just the recognised text, which throws away the visual record of the original document. For anything being submitted as proof of a certificate or identity, that is not an acceptable trade: the reviewer needs to see the document, not a transcription of it.

Where browser-side compression reaches its limit

Everything a browser can do to a scanned PDF operates on the container: deduplicating fonts, removing unused objects, compressing streams, stripping metadata. On a scan, that is a small fraction of the file and the result is a few percent. Genuinely shrinking a scan means decoding every embedded page image and re-encoding it at lower quality or resolution — heavy work that needs an image-aware pipeline running server-side. Being told this plainly is more useful than a tool that runs for thirty seconds and hands back a file 3% smaller without explaining why.

Frequently asked questions

What DPI should I scan documents at?: 200 DPI is legible for printed text and keeps files small; 300 DPI is a safe default when a reviewer may need fine detail. 600 DPI is for archival photographic reproduction and is rarely justified for a document upload.
Should I scan in colour or greyscale?: Greyscale for ordinary dark-text-on-white pages — it cuts the data to roughly a third with no meaningful loss. Colour only when a seal, stamp, ink colour or photograph matters for verification, which on official certificates it often does.
Does OCR make a scanned PDF smaller?: No, slightly larger. OCR adds a text layer on top of the existing page images. It makes the document searchable and copyable, which is valuable, but it is not a size-reduction technique.
Is a phone scan good enough for official uploads?: Usually yes, if you use a scanning app that crops to the page edges and corrects perspective, and you shoot in even indirect light. A skewed, shadowed photo of a page is rejected for legibility regardless of its resolution.
Why is my scanned PDF so much bigger than a Word export?: Because it stores a photograph of every page rather than the characters on it. A page of text is a few kilobytes as characters and several hundred kilobytes as an image of those same characters.
Can I convert a scanned PDF to text to make it smaller?: Technically yes via OCR, but you would be discarding the image of the original document. For proof-of-identity or certificate uploads that defeats the purpose, since the reviewer needs to see the document itself.
Does black-and-white scanning hurt legibility?: Not for clean printed text, where it is often the sharpest option. It handles photographs, shading and faint pencil or carbon-copy text poorly, so check the page content before choosing it.