Learn
PDF compression, explained without the marketing
Five short reads covering privacy, mechanics, and the situations where compression is the wrong move.
Why Online PDF Compressors Are a Privacy Risk
Drag a PDF onto a website, wait three seconds, download a smaller file. It feels harmless. But think about what actually happened: you took a document — maybe a resume with your address and ID number, a contract with pricing you don't want competitors to see, a medical form, a client's financials — and you handed a complete copy to a server you've never seen, run by a company you couldn't name.
Most free compression sites claim files are "deleted after 1 hour" or "processed securely." You have no way to verify that. You can't audit their servers, read their access logs, or confirm whether a backup job copied your document into cold storage. What you can see is that many of these pages load a dozen analytics and ad scripts, any of which can fingerprint the session. Some of these sites exist purely to harvest documents.
The risk math is simple. If you compress a flyer for a garage sale, the exposure is zero. If you compress a lease, a tax form, or an HR document, one retention mistake on their side means your data lives on infrastructure you don't control. For teams under GDPR, HIPAA, or a client NDA, uploading to a third-party processor can itself be a compliance violation — regardless of whether the file ever leaks.
The alternative is processing that happens where the file already sits: on your machine. This tool reads your PDF with the browser's own file APIs, re-encodes the images inside it, and hands you a new file. The bytes never touch the network. That's not a marketing claim; it's how client-side code works. You can unplug your ethernet cable halfway through and the job still finishes.
My rule after years in IT: never upload a document to a web tool unless you'd be comfortable posting that document publicly.
How PDF Image Resampling Works
A PDF is a container, not a single format with one compression dial. The text and vector shapes inside it are usually tiny — a page of crisp black type might weigh 5KB. The weight almost always comes from images: photos, logos, and especially scanned pages, which are really just photographs of paper.
When a PDF balloons in size, it's usually because someone exported it from a design program at maximum quality, or scanned it at 300–600 DPI when 150 would have looked identical on screen. Inside the file, those images live as compressed streams — most often JPEG, labeled DCTDecode in PDF internals.
Resampling means taking each of those image streams, decoding it back into pixels, and re-encoding it more aggressively. Two levers do the work:
- Quality. JPEG quality controls how much detail the compressor throws away. Quality 60 keeps a document perfectly readable and often halves the image bytes. Quality 30–40 is aggressive, aimed at job portals with hard limits.
- Pixel dimensions. A 300 DPI scan displayed at page size carries four times the pixels of the same scan at 150 DPI. Shedding pixels is the single biggest saving available, and the one most people never notice on screen.
This tool does exactly that, entirely in your browser: it locates the JPEG streams inside your PDF, pulls them through the canvas API, re-encodes them at your chosen quality and scale, and stitches the PDF back together with a fresh index. Text and vector art pass through untouched, which is why your fonts stay sharp.
One honest caveat: if your PDF's images are already heavily compressed, stored as JPEG2000, or scanned with a specialized monochrome method like CCITT, the browser can't reprocess them — and there's usually little to gain anyway. The tool tells you that's the case instead of pretending otherwise.
Email Attachment Limits in 2025
The limits haven't moved much, but the confusion around them has. Here's the working picture:
- Gmail: 25 MB per message, attachments included.
- Outlook.com: 20–25 MB depending on account type.
- Yahoo Mail: 25 MB.
- Corporate mail: often 10–15 MB, set by the mail gateway rather than the provider. The recipient's Exchange or Workspace admin may reject anything larger, and you'll only find out from a bounce message.
Then there's the inflation problem. Attachments get encoded into text for transit, which adds roughly a third to their size. A 20 MB PDF becomes about 27 MB once encoded — over Gmail's cap, even though the file "is" under it. Practical rule: if the provider says 25 MB, keep actual files under about 18 MB to be safe.
The real pain points:
- Job portals: applicant tracking systems routinely cap resumes and portfolios at 2 MB each.
- Government portals: commonly 5 MB per upload, sometimes per form field.
- Client email: nobody says anything, but a 15 MB attachment crawls on mobile data and trips spam filters more often than a lean one.
Targets that cover nearly everything: 2 MB for job applications, 5 MB for forms, 10 MB for everyday email. That's exactly why the presets at the top of this page exist with those numbers in their names.
If compression can't get you there — a photo-heavy portfolio, for example — the honest answer is a share link from a drive service rather than forcing quality down to mush. Compress first; link second. Don't email a 40 MB file and hope.
Compressing Scanned PDFs
Scanned PDFs are the heaviest documents most people own, and the most wasteful. Your scanner defaults to whatever setting it shipped with — usually 300 DPI color, sometimes 600 — and every page becomes a several-megabyte photo of a piece of paper. A 30-page lease can easily hit 90 MB even though it's nothing but black text on white.
Three settings fix this, in order of impact:
- Grayscale. If the source is black text on white, color information is noise. Converting scans from color to grayscale typically cuts image data by two-thirds before you touch anything else. Keep color only when the document genuinely needs it — signed stamps, highlighted markup, photos.
- Resolution. 150 DPI is the sweet spot for documents meant to be read on screen. Text stays sharp, signatures stay legible, and you're carrying a quarter of the pixels of a 300 DPI scan. Save 300 DPI for photographs or documents you plan to print at high quality again.
- JPEG quality. 50–60 is plenty for documents. Below 40, letter edges start to shimmer and small print in footnotes gets muddy.
The workflow that works: scan the document, run it through this tool with the Scanned Document preset, then open the result and zoom to 100% on the page with the smallest text you need to read. If it's legible, you're done. If not, bump the DPI one step and run it again.
One warning: if you ever need to search or copy text from the scan, run OCR before you compress aggressively. Compressed files stay readable for humans, but OCR engines prefer cleaner source images than your eyes do.
When NOT to Compress a PDF
Compression isn't free. It trades bytes for something else, and sometimes that trade is a bad one. Five situations where I tell people to leave the file alone:
- Digitally signed documents. A PDF signature is a cryptographic seal over the file's exact bytes. Recompress one image and the seal breaks — the signature shows as invalid or missing. If someone signed it, archive it as-is.
- Archival records (PDF/A). If a document was produced for long-term legal or regulatory retention, re-encoding it can knock it out of compliance. Courts and auditors care about this more than you'd think.
- Evidence. Anything that might end up in a dispute — incident reports, contract notices, insurance claims. You want the original bytes, timestamps intact, not a reprocessed derivative.
- Files that are already small. If the PDF is 800 KB, you might squeeze out 200 KB at the cost of image quality. Nobody's mail gateway cares about 800 KB. Skip it.
- Fine line art and technical drawings. JPEG recompression is brutal on thin lines, dimension text, and circuit traces. The artifacts collect exactly where precision matters. If the document is diagrams, maps, or CAD output, keep it lossless.
Also watch forms with embedded JavaScript or XFA: most survive recompression fine, but occasionally a form generator stores critical data in odd corners, and no browser tool can test every form engine on earth.
The pattern: compression is for transport copies — the version you email, upload, or stash in light storage. The master copy stays untouched on your drive. Make that habit stick and this tool can never hurt you.
