Compress a PDF Under 500 KB Without Blurry Text

A born-digital PDF — exported from Word, Pages, Google Docs or LaTeX — almost always fits under 500 KB already, because its text is stored as characters and font references, not pixels. A scan fits only at 150 DPI greyscale, or better, 1-bit black and white; a colour photograph or colour ID scan usually will not fit without damage you can see.

On a phone, Convertlix for Android (8.0 or later) re-encodes those images on the device: the PDF is opened through Android's own file picker, the work runs locally, and the result lands in Downloads. Nothing is uploaded — its network calls carry sign-in, licence, quota and payment traffic, never a document — so there is no server copy to retain. Sign-in is required once, with no guest mode; after that the tools run offline, free-tier files are capped at 25 MB, and the monthly conversion allowance is server-set and shown in the app.

Why 500 KB is harder than 2 MB

You are not imagining this: a file that dropped under 2 MB in one pass will bounce off 500 KB three times. 2 MB is generous: eight pages of 150 DPI greyscale land there unnoticed. 500 KB is a quarter of that, and the last quarter is where the trade stops being invisible. If your real cap is larger, take the gentler route in Compress a PDF Under 2 MB on Android (Free).

You can fix colour depth, resolution, and whether the pages are stored as pictures at all. You cannot make a photograph compress to 60 KB and still look like a photograph — arithmetic, not a tool limitation. When a portal wants a colour ID scan and 500 KB, one of the two will be approximate.

The arithmetic: 500 KB divided by your page count

The per-page budget decides which encodings are in play.

An A4 page scanned at 300 DPI is 2,480 × 3,508 pixels — 8.7 megapixels, a photograph of a sheet of paper. Halving the DPI to 150 quarters that to 2.2 megapixels, which is why resolution is the first lever and why Why Is My Scanned PDF So Large? is worth reading first. Per A4 page of typed text:

• Colour JPEG, 300 DPI: 1–3 MB — the scanner default.
• Colour JPEG, 150 DPI: 250–600 KB.
• Greyscale JPEG, 150 DPI: 100–250 KB.
• Greyscale JPEG, 100 DPI: 40–100 KB.
• CCITT Group 4, 1-bit, 300 DPI: 30–60 KB.
• JBIG2, 1-bit, 300 DPI: often 10–30 KB.

Past about four pages, nothing but bilevel fits the budget — and bilevel keeps letters sharpest. One caution: JBIG2's lossy symbol mode matches characters it judges identical, and implementations were documented in 2013 substituting similar digits, so where digits matter prefer CCITT Group 4 and read the numbers back.

What each approach costs you

Where the text actually goes blurry

A 10-point character is 10/72 of an inch tall: about 41 pixels at 300 DPI, 21 at 150, 14 at 100 and 10 at 72. Its strokes are a fraction of that — one or two pixels near 100 DPI. JPEG works on 8 x 8 blocks and discards high-frequency detail first, and a letter's hard black-to-white edge is high-frequency detail, so it returns as a grey halo. At 150 DPI the halo hides inside a two-or-three-pixel stroke; below about 100 DPI it is wider than the stroke, and that is your blur. Colour is worse: 4:2:0 chroma subsampling halves colour resolution in each direction, so coloured text smears first.

Bilevel has no gradient to ring, so edges stay hard even at 20 KB a page. The cost is total: photographs become dither patterns, a grey seal smudges or vanishes, faint pencil thresholds to white.

How to get a PDF under 500 KB

1. Find out which kind of PDF you have. Try to select a line of text. If words select, the size is one embedded object — a font family, an oversized logo, a pasted screenshot. If nothing selects, every page is an image.
2. Divide 500 KB by the number of pages. That tells you whether to aim for greyscale at 150 DPI or black and white at 300.
3. Go back to the source first. Re-export the Word or Pages original with "reduce file size" enabled, or rescan the paper at 150 DPI greyscale or in the scanner's black-and-white "text" mode — a driver that never creates the pixels beats a compressor that throws them away.
4. Drop colour depth before resolution. Greyscale first, 1-bit if the page is genuinely black ink on white paper. Keep colour only where a reviewer has to verify something coloured.
5. Re-encode the images inside the PDF. In Convertlix the tool is Compress PDF, which shrinks a PDF by re-encoding the images it contains; there is no target-size box, so treat it as one pass and check the size of the file it writes to Downloads.
6. If it is still over, go back to the original — never compress twice. Rescan lower, crop the margins, or drop pages you were not asked for. Then read every page.

When 500 KB is the wrong target

• A colour scan of an ID document. Passports and driving licences carry fine security printing a reviewer may be checking, and a rejection for illegibility costs more than a second attempt.
• Forms with stamps, seals or wet signatures. Bilevel destroys a grey seal, greyscale a coloured one; if the mark is the point of the document, protect it and negotiate on page count.
• More than about eight to ten scanned pages. Under 62 KB a page you are committed to black and white: fine for a typed contract, wrong for anything with a photo or a chart.
• A portal that also wants PDF/A. The goals conflict — PDF/A requires fully embedded fonts and forbids some compression schemes — so convert last and measure again. Convertlix does not produce PDF/A; LibreOffice exports it free from its PDF dialog, and Adobe's own export is the reference implementation.
• A cap that is not really 500 KB. For a visa or job application, check the real number before you cut anything; Get a PDF Under 10 MB for a Visa or Job Portal covers that case.

FAQ

Why does my 2-page PDF refuse to go under 500 KB?

Because two pages of colour scan at 300 DPI are two 8.7-megapixel photographs, and 250 KB per page is not room for a photograph of a page. Compressing that file again adds artefacts for maybe ten percent, which is why it feels stuck. Convert to greyscale at 150 DPI, or 1-bit black and white if the pages are plain black text, and the file usually lands between 60 KB and 500 KB.

Is 100 DPI readable?

On screen, usually yes for body text at 10 pt and above: 100 DPI renders a 10-point character about 14 pixels tall, legible but soft at the edges. It fails first on small print such as footnotes and reference numbers, and badly in print, because a printer asks for detail that is no longer there. 150 DPI is the lowest resolution that reliably survives printing.

Can I do this without installing anything?

Not really, unless you count what is already on the machine. macOS Preview has a built-in "Reduce File Size" Quartz filter that needs no install. Ghostscript is free and excellent from a terminal — gs -sDEVICE=pdfwrite -dPDFSETTINGS=/screen -o out.pdf in.pdf gets most scans under 500 KB, though /screen downsamples to 72 DPI, the setting that blurs scanned text, so try /ebook (150 DPI) first. Browser tools upload the file to their servers and meter their free tiers; check the site's own pricing page for the current cap (as of September 2026).

Does Convertlix upload my file?

No. The PDF is opened through Android's own file picker, re-encoded on the phone and written back to Downloads, so no copy is sent anywhere and nothing sits on a server to be deleted. Its network calls carry sign-in, licence, quota and payment data only, and it asks for no storage permission. The honest test is airplane mode: turn it on and the conversion still runs.

---

If the document started life on paper, the cheapest 500 KB you will ever save is a rescan at 150 DPI greyscale.