Why is my PDF so big? Scans, colour and DPI, explained
A typed document saved as a PDF is a few kilobytes. A scanned one can be several megabytes — per page. Both files end in .pdf, but they are built from completely different things. Here is where the weight actually lives, in plain words, and which parts of it you can take out without hurting the document.
A PDF of text is tiny. A PDF of a scan is a photo album
When you save a typed document as a PDF, the file stores letters and instructions: this character, in this font, at this position. A page of that weighs almost nothing — a few kilobytes — because a letter is a code, not a picture.
A scanner cannot do that. It does not read your document; it photographs it. The PDF it makes is a wrapper around that photograph, one image per page. So a "10-page PDF" from a scanner is really a photo album of 10 pictures — and everything about its size is the size of those pictures.
| The same page, four ways | Typical size | Why |
|---|---|---|
| Typed, saved as PDF | About 42 KB | Stored as text — letters are codes |
| Scanned at 150 DPI, greyscale | About 0.4 MB | A modest photo, one shade channel |
| Scanned at 300 DPI, colour | About 2 MB | Four times the pixels of 150 DPI, three colour channels |
| Scanned at 600 DPI, colour | 2 to 4 MB per page | The figure USCIS warns filers about — sixteen times the pixels of 150 DPI |
That last row is the one behind most portal failures on this site. USCIS's own filing guidance flags colour scans at high resolution running to several megabytes a page — and its cap is 12 MB per file. Four pages at 600 DPI and you are already at the wall.
The three things that add the weight
- DPI — the big one. DPI is dots per inch, counted in both directions. Double the DPI and you do not double the picture — you quadruple it: twice as wide and twice as tall. An A4 page at 300 DPI is about 8.7 million pixels; at 600 DPI it is about 34.8 million
- Colour. A colour pixel carries three channels of information; a greyscale pixel carries one. Where colour means nothing — a typed form, a statement — greyscale is the single biggest saving available
- How hard the picture is squeezed. The image inside the PDF is stored compressed, and the quality setting decides how much detail is kept. Scanners and phone apps often default to keeping nearly everything
Why your text PDF barely shrinks
This is the flip side people find suspicious: a compressor takes a 9 MB scan down to 900 KB, then barely moves a 300 KB text PDF. Nothing is broken. The compressor works on the pictures inside the file. A text PDF has almost no picture to squeeze — the letters were already stored the efficient way. A scan is nearly all picture, so it has nearly all the room to give.
Shrink it without hurting it
Step 1 — compress, aiming at the number. If a portal gave you a cap, do not guess with "low, medium, high". Compress PDF has a target-size mode that lands the file inside a window — the method is walked through in Compress a PDF to an exact size.
Step 2 — drop the colour where colour means nothing. Grayscale PDF converts a colour scan to shades of grey. Honest limit: where colour carries meaning — a degree's seal and stamps, a photo page — keep the colour and take the saving elsewhere.
Step 3 — remove what the scanner added. Crooked pages compress worse and read worse; Deskew PDF straightens them. And the truly blank backs a scanner adds to a single-sided document can come out with Remove Blank Pages — genuinely blank pages only; never delete a page a portal asked for, even one that looks empty.
Step 4 — next time, scan at the size documents need. For documents, 200 to 300 DPI is plenty; 600 DPI is for enlarging photographs. The camera-first way to scan is in Scan Documents with your phone. Every tool named here runs entirely in your browser — your files are never uploaded.
Take the weight out
Compress PDF shrinks the photograph inside your scan — and its target-size mode lands the file inside a portal's window, not just 'smaller'.
Open Compress PDFA worked example: four pages, straight into the wall
The table above stops being abstract the moment a real file meets a real cap. Say a filer scans a four-page document the way a scanner app suggests by default: colour, at 600 DPI. The table prices that choice at 2 to 4 MB per page — so the file arrives between 8 and 16 MB, and against USCIS's 12 MB per-file cap it may already be refused. Nothing about the document is unusual: the weight was decided entirely by two settings, the resolution and the colour.
Now run the same document through the shrink steps in order. The pages are typed forms, so colour carries nothing: Grayscale PDF removes two of every pixel's three channels. Compress PDF then re-renders the pictures at the size documents need — the 200 to 300 DPI band this page recommends — instead of the 600 DPI the scanner chose, and each halving of DPI quarters the pixels rather than merely halving them. The filer zoom-checks the smallest print, and uploads a file that sits comfortably inside the cap. The untouched original stays in the folder as the master. The portal-window method behind the last step is walked through in compress a PDF to an exact size.
Edge cases: the files that only partly shrink
A mixed PDF. Many real files are both kinds at once — typed pages saved from Word, with two scanned pages appended. Compression appears to “barely work” on such a file, because the text pages were already a few kilobytes each and had nothing to give; the saving all comes out of the scanned pages, which is where the megabytes were. The blank backs. A scanner run over a single-sided document adds a photographed blank page after every real one, doubling the album. Remove Blank Pages takes out genuinely blank, scanner-added pages — and only those; a page a portal asked for stays, even one that looks empty. Crooked pages. A page photographed at an angle compresses worse and reads worse; Deskew PDF straightens it.
Sources
- Adobe — Compress PDF files online: Adobe's own account of where PDF weight lives — embedded images first — and what its optimizer removes, matching the four-way page comparison in this guide.
- Tool behaviour — the image-resolution gates and greyscale step used in the worked example are from PDFJadoo's own Compress PDF page, checked 18 November 2026; the companion guide to compressing without losing quality covers the readability floor in depth.
Before you upload: the 60-second check
- You know which kind of PDF you have: text, scan, or a mix
- Scan resolution is sane — 200 to 300 DPI for documents
- Colour is kept only where it carries meaning (seals, stamps, photos)
- Only genuinely blank, scanner-added pages were removed
- The final size is checked against the portal's window — floor and cap
Fair questions
Why is a one-page PDF 5 MB?
Because that page is a photograph, not text. A colour scan at high DPI holds tens of millions of pixels; one page like that outweighs a whole typed document. The fix is to compress the picture inside it — or, next time, scan at 200 to 300 DPI.
Does compressing a PDF hurt the quality?
Compression works on the pictures inside the file. Typed text is stored as text and stays sharp. Photos and scans give up a little detail — at sensible settings the difference is invisible on a document page, and the 60-second zoom check tells you the truth before you upload.
What DPI should I scan documents at?
200 to 300 DPI for documents. Text stays crisp, faces stay recognisable, and the file stays a sane size. 600 DPI quadruples the pixels again and is meant for photographs you plan to enlarge — it is the setting behind most 'file too large' errors.
Why won't my text PDF compress any smaller?
There is almost nothing in it to compress. A text PDF stores letters as codes, which is already the smallest way to hold them. Compressors squeeze images; a file with no images has no room to give — that is a property of the file, not a fault in the tool.
Is greyscale safe for official documents?
It depends what the colour is doing. For typed forms and statements, colour carries nothing and greyscale is safe and much smaller. Where the colour is part of the proof — seals, stamps, a passport photo page — keep it, and take your saving from compression instead.
My PDF has ten pages but only two are scans. Why is it still so big?
Because the two scanned pages carry nearly all the weight. A typed page stored as text weighs a few kilobytes — the table above prices a whole typed page at about 42 KB — while one colour scan at high DPI is a photograph of millions of pixels. Compress the file and the saving comes out of those two pages; the text pages had almost nothing to give.
Will removing blank pages make a big difference to the size?
Only if your scanner added them. Genuinely blank, scanner-added backs are pure extra photographs, so removing them helps — but the real weight lives in resolution and colour, not page count.
Source for the 600 DPI figure: U.S. Citizenship and Immigration Services, Tips for Filing Forms Online (uscis.gov) — high-resolution colour scans run to several megabytes per page, against a 12 MB per-file cap. Pixel counts are plain arithmetic from the A4 page size. Checked 10 November 2026.
Keep reading
Compress PDF
Shrink the photograph inside your scan — target a window, not just 'smaller'.
Compress to an exact size
200 KB, 100 KB, any window — portals check numbers, so aim at the number.
Grayscale PDF
Where colour means nothing, dropping it is the biggest single saving.
Scan Documents with your phone
Camera first, sensible size from the start — no app to install.