Why Scanned PDFs Are So Big — and How to Compress Them
A scanned PDF can be large because each page contains an image, sometimes with an OCR text layer. Several high-resolution color pages can add up to tens of megabytes. In Compress PDF, start with 'Keep selectable text'. If needed, explicitly choose 'Compress as images', accepting loss of selectable text, search, links, interactive fields and possibly detail. Set a target and click 'Compress files', then review output bytes and every page. Processing stays in your browser. Below: why scans are so big and how to compress them well.
Why is a scanned PDF tens of megabytes?
A scan stores page images, sometimes alongside an OCR text layer. Image size depends on resolution, color and encoding: scanners often default to 300 or 600 DPI in full color, so one A4 page can be several megabytes and a dozen pages add up fast.
A text PDF exported straight from Word or Google Docs is the opposite — the text is vector data and barely adds size, so the same number of pages is often just a few hundred KB. So 'same ten pages, why is the scan the big one?' comes down entirely to one being images and the other being text.
- High DPI: scans default to 300–600 DPI; more resolution means more pixels and a bigger file.
- Color: a color scan is far larger than grayscale or black-and-white, and many text documents don't need color at all.
- Text layers: a scan may have no selectable text, or it may include OCR text; check before choosing image compression.
Choosing a compression method for scanned PDFs
The default 'Keep selectable text' method re-saves PDF structure and may barely shrink a scan. 'Compress as images' is an explicit choice for each file: it rebuilds pages as pictures and may reduce size, but removes any existing OCR layer, links and interactive fields and may soften detail. There is no fixed savings percentage.
- Open Compress PDF and drag the scan in (several at once is fine).
- Set a target in KiB and compare its displayed bytes with the upload limit. Choose the method for each scan; a target does not change the method automatically.
- Click 'Compress files' to process locally. If a file fails, the others can still finish; an output may remain above target.
- Compare the before/after, zoom in to confirm the text is still crisp, and download.
How do you shrink it without losing legibility?
The risk with scans is losing readability while trying to reduce size. Consider grayscale at the source when color is unnecessary, review each output, and remove pages only where submission rules allow. These steps do not guarantee a reachable target.
- Save size at the source: for black-and-white text, scan in 'black & white' or grayscale — much smaller, and steadier to compress.
- Tighten in steps: hit a looser target first, confirm it's acceptable, then push smaller — don't aim for 50KB straight away.
- Small targets may be difficult for many-page scans. If allowed, remove blanks with Delete Pages or keep only needed pages with Split PDF.
- Zoom in page by page afterwards to confirm text, stamps and signatures are clear before uploading or sending.
Frequently asked questions
Why is my scanned PDF tens of megabytes when it's only a few pages?
Each scanned page contains an image, sometimes with an OCR text layer too. A 300–600 DPI color scan can be several MB per page. Image compression may reduce that size, but savings and legibility depend on the source and settings.
Can I still select and search the text after compressing a scan?
An existing OCR layer can remain with 'Keep selectable text'. 'Compress as images' removes that layer and makes text unselectable and unsearchable. If the scan never had a text layer, compression does not add one; OCR is a separate step.
How should I scan so the file is smaller to begin with?
Scan in grayscale or black-and-white (color is unnecessary for plain text) and set the DPI to 200–300 (sharp without being bloated). Size you save at the source costs less clarity than crushing it afterwards.
Is compressing a scan safe — is it uploaded?
Safe, and not uploaded. Compression runs locally in your browser via WebAssembly, so scanned contracts, IDs and the like stay on your device and never touch a server.
Updated · compress cat team