Why scans get so large
The scanner turns every page into a grid of pixels. How many pixels is set by the resolution in dpi (dots per inch) — and it scales with the square: an A4 page at 300 dpi has about 2480 × 3508 pixels, 8.7 million. At 600 dpi it is 4960 × 7016, 34.8 million — four times as many for twice the resolution.
Then there is colour depth. In colour every pixel takes three bytes, in greyscale one, in pure black and white an eighth of that. Uncompressed, a colour A4 page at 300 dpi weighs 26 MB; the scanner's JPEG compression brings that down to one or two megabytes. As an order of magnitude for a single A4 page:
| Scanner setting | Pixels | Typical size per page |
|---|---|---|
| 200 dpi, greyscale | 1654 × 2339 | 150–300 KB |
| 300 dpi, greyscale | 2480 × 3508 | 300–600 KB |
| 300 dpi, colour | 2480 × 3508 | 0.8–2 MB |
| 600 dpi, colour | 4960 × 7016 | 3–8 MB |
| 300 dpi, black and white (1 bit) | 2480 × 3508 | 30–100 KB |
The numbers vary with the document and the scanner's compression; the ratios stay. Ten pages in colour at 600 dpi come to 30–80 MB — more than any portal accepts. The same ten pages in greyscale at 300 dpi weigh 3–6 MB and fit almost anywhere.
What happens when you shrink it
ReduPDF finds the page images inside the PDF, scales them down to a smaller edge length and re-saves them as JPEG. In the default setting the longest edge is limited to about 1400 pixels — roughly 120 dpi on A4. For a clerk's screen that is more than enough; digits, stamps and signatures stay legible. 300 dpi is a benchmark for print, not for display.
With a target size, ReduPDF goes only as far down as needed to stay under it, and no further. A 20 MB colour scan usually ends up at one or two megabytes without you having to know any settings.
Choose a PDF or photo
How small should the file get?
Without a target size, ReduPDF compresses in a balanced way. With a target size, it automatically compresses as much as needed to stay under it.
Limits the longest edge of the photo in pixels — independent of the target size.
Advanced settings
Only needed if you want to set a specific image quality instead of a target size.
Start shrinking
Setting up the scanner — for next time
200 to 300 dpi. Entirely sufficient for receipts, letters and contracts. 600 dpi adds nothing for text but costs four times the storage.
Greyscale instead of colour whenever the document carries no colour information — receipts, invoices, official letters. A coloured stamp is still recognisable as a stamp in grey.
Black and white only for pure text pages. The 1-bit mode is extremely compact, but faded thermal receipts, signatures and stamps fall apart into specks.
Document mode instead of photo mode. Many scanners and phone apps offer both; photo mode saves at camera resolution — twelve megapixels on a phone, three to five megabytes per page.
Save straight to PDF, not TIFF or PNG. Those formats compress scans poorly or not at all.
What happens to the OCR text
Scanners with text recognition lay an invisible text layer over the page image — that is why you can search and select text in such scans. In its first stage ReduPDF touches only the images; that text layer is kept completely intact.
Only when a chosen target size cannot be reached any other way does ReduPDF rebuild the pages entirely as images. Then the text layer is lost. The result tells you which case applied — and you can tell by whether text can still be selected in the shrunk PDF.
Frequently asked questions
How many dpi are enough for a receipt?
200 to 300 dpi. Digits, stamps and signatures stay cleanly legible. 600 dpi quadruples the file without a clerk seeing any difference on screen.
Colour or greyscale?
Greyscale, as long as the document does not need colour information. That cuts the data to a third, and a coloured stamp is still recognisable in grey. Colour only when the colour itself is content — ID cards or plans, for example.
What happens to the OCR text?
It is kept as long as only the page images are compressed. Only when ReduPDF has to rebuild pages entirely for a very small target size is the text layer lost — the result tells you which case applied.
Why won't my scan get any smaller?
Usually because it is already heavily compressed: 1-bit black-and-white scans are tiny to begin with, and strongly compressed JPEGs can hardly be squeezed further without becoming illegible. ReduPDF also leaves CMYK images untouched. In such cases a target size helps, which triggers the stronger stage — or splitting into several files.