How to Compress a PDF Without Losing Quality
Short answer: a scanned or photo-heavy PDF typically shrinks by 60–90% when every page is re-rendered at about 120 DPI and re-encoded: seconds of work, free, in your browser, nothing uploaded. A digital text PDF is a different animal: it is already compact, and forcing it through the same pipeline turns selectable text into pixels for a much smaller gain. Match the preset to the file you have: Medium for scans, High for fine text, reduce digital text at the source instead.
What actually happens when a browser compresses a PDF
A browser compressor does not squeeze your existing file the way an archive tool would. It rebuilds the document from scratch, one page at a time:
- Render. pdf.js (bundled in the page) draws each page onto an off-screen canvas at the preset resolution: 150, 120 or 96 DPI. What comes off that canvas is a bitmap, whatever the original page contained.
- Re-encode. The bitmap is compressed as a JPEG, which is where most of the savings live: a lossless or full-quality image becomes a lean, lossy one.
- Rebuild. pdf-lib assembles a brand-new PDF around those images with a clean object structure. Document metadata (title, author, the software string, timestamps) is dropped along the way, trimming kilobytes and a slice of private detail.
Every step runs locally: read from disk, rendered in the tab, re-encoded, downloaded. The file-handling policy spells the pipeline out in writing, and DevTools → Network confirms it while the tool runs.
The consequence is this article's whole story: the new pages are pictures of your document, not its text objects. Keep that in mind below.
Which preset to use
Three presets, one decision. The right choice depends on what is inside the file, not on how many megabytes it starts with.
| Preset | Pages rendered at | Reach for it when | Typical result |
|---|---|---|---|
| High: 150 DPI | finest detail | small text, dense tables, detailed diagrams, tiny print | least shrink, type stays crispest |
| Medium: 120 DPI (recommended) | balanced | ordinary scans, forms, mixed documents | scans usually come out 60–90% smaller |
| Low: 96 DPI | coarsest | smallest attachment, screen-only reading | smallest file; small text softens |
Text-heavy documents: use High. Fine print rendered at 96 DPI looks mushy on a phone and worse on paper, so if you compress a mostly-text file anyway, High (150 DPI) is the least bad setting. But even High rasterizes. You are choosing how sharp the picture is, not whether it stays text. For digital text PDFs, shrink the file where it was made instead: re-export with lower image quality, or "Print to PDF" with a minimum-size option. Adobe tiers its own online compression levels the same way, because every tool that renders pages faces this trade-off.
Step-by-step: how to compress a PDF
Everything below happens in the free PDF Compressor: no install, no signup, no file-size ceiling.
- Drop your files in. Drag a PDF onto the dropzone or click to browse. Batch is supported. Files queue and process one after another, so a folder of scans is one session rather than ten.
- Pick a preset. Medium (120 DPI) for scans, High (150 DPI) for small text, Low (96 DPI) when size is all that matters. Unsure? Start at Medium.
- Watch the per-page progress. Each page renders in turn, so a long document takes proportionally longer; the status line reports pages done and bytes saved so far.
- Compare before and after. Every row shows the original size, the new size and the pages processed, a figure measured on your own file rather than promised on a landing page.
- Download, or keep your original. If the rebuilt file would be larger than your original, the tool returns the original instead. Metadata is stripped either way, and a re-run with another preset takes one click.
Everything runs on your device, so contracts and scanned IDs never leave it, and the tool works offline once loaded.
Why your PDF won't get smaller
If the number barely moved (or went up), one of these six is usually why.
- It is already optimized. A file already run through a PDF writer at maximum compression has little slack left. Fix: try Low once; if the size still refuses to drop, go back to the source and export at lower image quality.
- It is mostly text. Text already compresses extremely well inside a PDF, so re-encoding it as JPEG often adds weight rather than removing it. Fix: leave it alone, or reduce it at the source with a minimum-size print.
- It is vector-heavy. Logos, charts and line drawings are coordinate data, not pixels. Flattening them into images can cost more than it saves. Fix: simplify the artwork, or lower the quality of embedded rasters.
- Fonts are embedded in full. A whole typeface instead of a subset can add hundreds of kilobytes. Fix: "Print to PDF" often rebuilds a leaner copy; desktop optimizers can unembed or subset fonts. Adobe's advanced size-reduction guide covers this.
- It is a scan saved wastefully. Low resolution does not guarantee a small file: pages stored as uncompressed bitmaps or full-quality JPEGs stay heavy at any DPI. Fix: run Low once: re-encoding strips that padding.
- You are comparing the wrong numbers. The "document size" dialog shows page size, not file size, and stripping metadata also shifts the number you remember. Fix: compare the file on disk (Properties on Windows, Get Info on a Mac) against the before/after figures on each row.
And when the file is simply enormous (300 pages of catalog, a year of statements), the cleanest reduction is fewer pages. Split the PDF, send the part that matters, and the problem disappears before compression starts.
The honest trade-off: searchable text vs. size
This is the part most compressor pages skip, and it is why people end up disappointed with good tools.
When a page is re-rendered as an image, everything in its text layer goes with it: selecting a sentence, Ctrl+F, copy-paste, a search index matching a clause. The page still looks the same, so you never see what you lost.
So split your files into two piles:
- Text matters → do not rasterize. Contracts, papers, manuals (anything you will search or quote later): keep the text PDF as it is, or cut its size at the source. A 4 MB searchable file beats a 1 MB picture of the same words the moment you need one back.
- Only the picture matters → rasterize freely. Scans, photo albums, signed forms you will read once on a phone: here compression is a straight win. The text layer you are "losing" was never there.
We would rather you keep a larger file that works the way you need than shrink it into something you cannot search. Choose the preset with the trade-off in view.
Compressing a PDF for email
Attachment limits are why most people search for this. Thresholds vary by provider and employer, so treat these as typical targets, and check the number on your own file:
- Under 5 MB: usually trivial. Most consumer mail providers allow far more, and a typical scan at Medium (120 DPI) lands inside the limit first time.
- Under 1 MB: typically needs Medium, and Low (96 DPI) for long or image-dense documents. Still over the line after a Low run? The pages, not the pixels, are the problem.
- Under 500 KB: plan on Low plus fewer pages: halving the page count halves the weight in a way no DPI setting can match.
One habit worth keeping: compress, read the printed size, then attach. A re-run at the next-lower preset takes seconds. The same arithmetic applies to images: see how to compress a photo to 20KB, 50KB or 100KB for online forms.
Password-protected and huge PDFs
Encrypted files cannot be compressed in the browser. The renderer has to draw the pages, and an encrypted PDF cannot be drawn without its password. There is nothing to work on. The fix is simple: open the file in a reader where you know the password, remove the security, or use "Print to PDF" to create an unlocked copy, then compress that. No password, no compression, which is a feature rather than a bug.
Very large files are a memory question. Every page renders at full resolution in the tab, so hundreds of pages can exhaust a phone's memory before any file-size limit does. If compression fails or the browser asks to reload, close other tabs and retry on a desktop browser, or split the document and process it in parts. A PDF Splitter run takes seconds. Typical documents never hit this wall; 400-page scans occasionally do.
Quick FAQ
Does compressing a PDF ruin its quality?
Not usually, but it changes what the file is. A scan at Medium (120 DPI) looks the same while shrinking by roughly 60–90%. What you trade is the text layer: fine print is best served by High (150 DPI). The honest word is trade, not ruin.
Why is my PDF file so large?
Usually because it is a scan or a photo-heavy export. One full-color page at high resolution can outweigh a hundred pages of text. Embedded fonts and maximum-quality exports add weight too, and an already-optimized file has little left to remove.
How can I compress a PDF under 1 MB?
Start with Medium (120 DPI) on scans, then Low (96 DPI) if you are still over. If it still will not fit, split it and send only the part that matters. Fewer pages shrinks an attachment faster than any preset, though exact sizes depend on your file.
Do I have to install anything?
No. The compressor runs in the browser tab you are already in: open it, drop the file, download the result. No desktop app, no account, no extension, and once the engine has loaded it keeps working offline.
Is my PDF uploaded?
No. Reading, rendering, re-encoding and downloading all happen on your device with pdf.js and pdf-lib, so no server ever sees the file. Verify it: open DevTools, switch to Network and compress a document while watching nothing go out.
Can I compress a password-protected PDF?
Not directly. The browser cannot render an encrypted PDF without its password, so there is nothing to re-render. Open the file in a reader, remove the security or use Print to PDF to make an unlocked copy, then compress that.