Why scanned PDFs are huge: the DPI and colour math

The raw math of 1 scanned page

A scanner measures the page in dots per inch (dpi) and stores 1 pixel per dot. An A4 sheet is 8.27 by 11.69 inches, so:

• at 150 dpi it becomes 1240 by 1754 pixels: 2.2 million pixels
• at 300 dpi, 2480 by 3508 pixels: 8.7 million pixels
• at 600 dpi, 4960 by 7016 pixels: 34.8 million pixels

Each pixel then costs a number of bits that depends on the colour mode: 24 bits (3 bytes) in colour, 8 bits (1 byte) in grayscale, and 1 bit in black and white. Multiply, and you get the size of the page before any compression:

A4 page	Colour	Grayscale	Black and white (1-bit)
150 dpi	6.2 MB	2.1 MB	266 KB
300 dpi	24.9 MB	8.3 MB	1.0 MB
600 dpi	99.6 MB	33.2 MB	4.1 MB

(MB here is 1,048,576 bytes, the way Windows and our result readout count.)

Nobody stores pages raw; scanners compress them, usually as JPEG. Our 10-page test lease, scanned at 300 dpi in colour and saved at the JPEG quality a scanner app uses, comes out at about 1.8 MB a page and 17.6 MB in total. That is a 14 times saving on the raw 24.9 MB, and it is still too big for most portals and many mail servers.

Why doubling the dpi quadruples the file

Resolution counts in 2 directions at once. Doubling the dpi doubles the pixels across and doubles them down, so 600 dpi holds 4 times the pixels of 300 dpi, not 2 times. Halving it works the same way in reverse: 150 dpi holds a quarter of the pixels of 300.

That squared rule is why resolution is the strongest lever on a scan’s size. We measured it on page 1 of our test lease (1.7 MB as scanned), re-encoded at the same JPEG quality (75) each time:

Resolution	1 page
300 dpi	1.0 MB
200 dpi	533 KB
150 dpi	330 KB
100 dpi	167 KB
75 dpi	100 KB

From 300 to 150 dpi the page shrinks 3 times. That is less than the 4 times the pixel count predicts, because a smaller picture packs more detail into each pixel, but no other setting comes close. Reduce PDF DPI shows what each step looks like and which to pick.

Where each step is visible: 300 dpi keeps 6-point footnotes and stamps crisp at 200 percent zoom. 150 dpi reads cleanly at page width on a laptop and when zoomed a little. 110 dpi is fine for normal-sized text on a screen. Below about 75 dpi, small print starts to break up.

Colour costs more raw than it does after JPEG

Raw, colour costs 3 times grayscale and 24 times black and white. Those ratios shrink once JPEG is involved, because JPEG stores colour detail at reduced resolution and spends few bytes on the colour of plain paper.

On our 10-page test scan, converting to grayscale on its own took the file from 17.6 MB to 15.4 MB: 12 percent smaller. Grayscale followed by the Email preset gave 1.7 MB, the same as the Email preset on the colour file. On page 1 alone, grayscale then Email gave 172 KB against 177 KB in colour. So on scans of black-on-white paper, the colour mode matters far less than the resolution, and grayscale is a small extra step rather than the main one.

Black and white (1-bit) is a different story. A clean 1-bit scan of a typed page is very small, but it turns grey areas, photos and faint pencil into hard-edged dots, and it cannot be undone. Our compressor leaves 1-bit images exactly as they are; they are already compact.

Why 2 scanners give different sizes for the same page

Scan the same letter on 2 devices and you can get 400 KB on one and 6 MB on the other. The page did not change; the settings did:

1. Resolution. Office copiers often default to 200 or 300 dpi, phone scanner apps pick their own, and “best quality” modes go to 600.
2. Colour mode. Some devices default to colour, others to grayscale or black and white for documents.
3. Compression quality. The same JPEG at a high quality setting can be twice the size of one at a moderate setting.
4. Format inside the PDF. Some scanners store pages losslessly, which is far bigger than JPEG for paper grain. Others split the page into a sharp black-and-white text layer and a soft background, which is very small.

You can see which kind you have: open the file in the inspector and read the image formats. DCTDecode is JPEG, FlateDecode is lossless, CCITTFaxDecode and JBIG2Decode are 1-bit.

Shrinking a scan after the fact

You do not need to scan again. The compressor decodes each page picture, shrinks it to the preset’s resolution and re-encodes it as JPEG. On the 10-page test scan:

Setting	Result	Per page
As scanned (300 dpi)	17.6 MB	about 1.8 MB
Print preset (300 dpi)	unchanged, and the tool says so
Web preset (150 dpi)	3.2 MB	about 320 KB
Email preset (110 dpi)	1.7 MB	about 180 KB

The Print preset returns this file unchanged because its pictures are already at 300 dpi; re-encoding them would not make them clearly smaller, and a result that is not smaller is never delivered. The full step-by-step, with size targets for portals, is in Compress a scanned PDF.

What this cannot fix

Lost detail does not come back. Once a page is at 110 dpi, no tool can restore what 300 dpi held. Keep the original scan and send the small copy.

Fine print has a floor. 150 dpi grayscale reads fine on screen for normal text, but 6-point footnotes, faint stamps and handwriting in pencil may need 300. When the receiver must read those, send fewer pages at higher quality rather than every page crushed.

A smaller scan is still a picture. Compression does not make the text searchable or selectable. For that, run OCR on the scan while it is still sharp, then compress the result; OCR reads a 300 dpi page far better than a 110 dpi one.

The best fix is at the scanner. If you scan often, set the device to 150 or 200 dpi and grayscale for paperwork. Every file after that starts small.