How to Compress a PDF Without Losing Quality

Where do PDF megabytes actually come from?

A PDF is not one picture — it is a container holding text, vector graphics, embedded fonts, images and metadata. The text part is tiny. Everything heavy is visual, and in the wrong combination it adds up fast:

The pattern is clear: if you fix the images, you have fixed the file. Text compresses so well that a 200-page typed report is often under 1 MB — if yours is 40 MB, blame the pictures. This is also why the free PDF compressor focuses on the image pipeline: that is where 90% of the savings live.

What "without losing quality" really means

There are two kinds of quality here, and they are not the same. Lossless means every pixel is bit-for-bit identical to the original — ZIP-style compression on the images. It is safe but rarely saves much, because the images inside most PDFs are already compressed. Visually lossless means you discard data your eyes cannot see — fine grain, imperceptible color steps — and the result looks identical even though the bytes changed.

Visually lossless is what makes 10–20× shrinkage possible, and it is honest: a 600 DPI scan viewed on a 96 DPI screen throws away most of its pixels at display time anyway. Removing them beforehand changes nothing you can see. The trick is knowing exactly how far to go — which is what the DPI table below is for.

The 5 techniques that shrink a PDF

Good compressors combine these, from biggest effect to smallest:

1. Downsample images to the target resolution. A scan captured at 600 DPI contains 16× the pixels of the same page at 150 DPI (4× per dimension, squared). If the document will be read on screen, those pixels never do any work. Resize once, at compression time, and the bulk of the size vanishes.
2. Recompress with sensible quality. JPEG quality 90+ is near-lossless and huge; quality 70–85 keeps photos and scans looking pristine at a fraction of the size. Below 60, blocky artifacts creep into text edges and gradients.
3. Deduplicate repeated images. Company logos, letterheads and watermarks appear on every page but are often embedded separately per page. One copy referenced 20 times beats 20 copies.
4. Subset embedded fonts. Embedding a full 600 KB font for a document that uses twelve characters is wasteful. Subsetting embeds only the glyphs used, often shrinking the font to a few kilobytes.
5. Strip the invisible extras. Document metadata, preview thumbnails, editing history and unused objects accumulate in files passed around for years. Removing them only saves 1–3%, but it costs nothing.

Notice what is not on the list: flattening transparency, rasterizing text, or deleting the OCR layer. Those destroy functionality (searchability, selectability, print fidelity) in exchange for modest gains — a real quality-preserving compressor never touches them by default.

How small should the images be? The DPI rule

DPI — dots per inch — decides how many pixels a page holds. Pick it by destination, not by habit:

The most common mistake is leaving the scanner at its 600 DPI default and never touching it again. A quick rule of thumb: if nobody will print this document, 150 DPI is generous. Try the PDF compressor at 150 DPI first — if the zoomed-in text still looks crisp, you are done; if not, nudge it to 200.

A worked example: 48 MB scan down to ~3 MB

Take a real-world case: a 10-page rental contract scanned in color at 600 DPI, JPEG quality 95, saved as PDF — file size 48 MB, too big for the leasing portal's 10 MB upload limit.

Step 1 — count the pixels. An A4 page at 600 DPI is about 4,960 × 7,016 ≈ 34.8 megapixels. JPEG quality 95 stores roughly 1.4 MB per megapixel for photographic content, so each page is ~4.8 MB and ten pages give us our 48 MB. Checks out.

Step 2 — downsample to 150 DPI. 600 ÷ 150 = 4× fewer pixels per side, so 16× fewer total: 34.8 ÷ 16 ≈ 2.2 megapixels per page. At the same quality, the images alone drop from 48 MB to ~3 MB.

Step 3 — recompress at quality 80. Dropping JPEG quality from 95 to 80 roughly halves photographic file size with no visible difference at reading distance: ~3 MB → ~1.5 MB for the images, plus the PDF container overhead.

Result: ~2–3 MB — comfortably under the portal limit. Opened side by side with the original at normal zoom, the two are indistinguishable: letters are sharp, the signature is intact, and the OCR text layer was never touched, so the document still searches and copies exactly as before.

When compression will visibly hurt (and how to avoid it)

Three pitfalls account for nearly every "my PDF looks worse now" complaint. First, double compression: recompressing an image that is already heavily compressed (a screenshot pasted into Word, exported at low quality) multiplies artifacts. Always compress from the highest-quality source you have. Second, over-downsampling small text: footnotes and fine print below 8 pt get soft under 150 DPI — keep those pages at 200. Third, line art as JPEG: diagrams and charts with sharp edges compress badly as JPEG; if your compressor offers it, keep vector graphics and line drawings lossless and only recompress photographs.

One safeguard covers all three: always keep the original until you have opened the compressed copy, zoomed to 150%, and confirmed it looks right. Storage is cheap; rescanning a signed contract is not.

Practical tips

• Fix the scanner first. Scanning at 150–200 DPI directly is better than scanning at 600 and compressing later — the file is small from birth and you skip a step.
• Grayscale beats color for text. A black-and-white contract scanned in grayscale is roughly a third the size of the color version, and nobody misses the color.
• Compress before you sign, then check. If the PDF needs a digital signature, compress first and sign last — some signature workflows reject files modified after signing.
• Mind the upload limit math. Portals usually mean 10.0 MB, not "about 10" — aim for 8 MB max to leave margin for their size calculation.
• Split when compression is not enough. If a 300-page report will not fit any limit, split it into chapters — the PDF splitter does it in your browser — and compress each part.

Frequently asked questions

Will compressing a PDF make my images blurry?

Only if you push it too far. Images keep their full visible sharpness as long as you stay at or above the resolution your reader actually needs — 150 DPI for screens, 200–300 DPI for print — and keep JPEG quality at 70–85. Artifacts appear when you recompress an already-compressed image at a much lower quality or downsample below the display resolution.

Why do scanned PDFs shrink so much when compressed?

Scanners capture pages as images at very high resolution — often 600 DPI — which is far more pixels than any screen or most printers need. A single scanned page can hold 10–25 megapixels. Downsampling those images to 150 DPI throws away 16 out of every 17 pixels with no visible difference on screen, so a 50 MB scan commonly drops to 3–5 MB.

Can I compress a PDF that is mostly text?

There is little to gain. Born-digital PDFs — documents typed in Word and exported — are usually only a few hundred kilobytes already, because text and vector graphics compress almost perfectly. If such a file is still large, the bulk is almost certainly embedded images, so apply the same image techniques described above and re-check.

Is it safe to upload my PDF to a free online compressor?

It depends on the document. Many free compressors upload your file to their servers, which is a poor idea for contracts, IDs, medical records or anything with personal data. A safer route is a tool that processes entirely in your browser, like the DailyWebKit PDF compressor — the file never leaves your computer, so there is nothing to leak, log or breach.

How small can I make a 100 MB scanned PDF?

Typically 5–10 MB with no visible loss, using 150 DPI downsampling and JPEG quality around 75–80 for the page images. Aggressive settings — 100 DPI and lower quality — can push it under 3 MB, but fine print starts to look soft when zoomed. For anything that will be printed, keep 200–300 DPI and accept the larger file.

Does compressing a PDF remove the text layer or annotations?

A careful compressor only touches images and leaves text, links and the OCR layer intact. Problems arise with careless settings: flattening can merge annotations into the page, and some tools strip attachments, bookmarks or form fields. Before compressing an important document, check the options — and always keep the original file until you have verified the compressed copy.

Related tools