What the PDF Compressor actually does
Most PDF compressors online — iLovePDF, SmallPDF, Adobe's web compressor — promise a smaller file and don't explain the trade. Our PDF Compressor does one specific thing: it rasterizes every page. That means it renders each page as an image, then writes a new PDF that contains those images. Big size drops on the right input. Real losses on the wrong input. We tell you which one you have before you click.
This is the opposite of how "real" PDF compression works. A proper PDF optimizer recompresses the embedded images and fonts inside the file while keeping the text as text. That kind of compressor is what Adobe Acrobat does on the desktop, and what iLovePDF does on a server after you upload your file. We don't do that yet — it requires either a server (and your file leaving your machine) or a heavier WASM toolchain we haven't shipped. So instead of pretending, we ship the honest version: rasterize, warn, let you decide.
When this tool is the right answer
Image-heavy PDFs are where rasterize-and-recompress shines. A 50 MB scan of a 30-page contract typically drops to 3-6 MB at the Balanced preset, no visible quality loss. A photo album exported from your phone might shrink from 80 MB to 8 MB. Screenshot collections, scanned receipts, real-estate listings, signed paperwork — anything where the page is already a picture — compresses dramatically because the original JPGs were saved at higher quality than they need to be.
The math is straightforward. A scanned page stored at 300 DPI as a full-quality JPG is roughly 1-3 MB. Re-rendered at 100 DPI and re-encoded at 75% JPG quality, the same page is 100-300 KB. Across a 30-page document that's the difference between an email attachment that gets bounced and one that goes through.
Honest warning: if your PDF is text-heavy — a book, a contract you'll search inside, a research paper, anything where you need to copy quotes or search for words — don't use this tool. You will lose selectable text, embedded fonts, and the ability to Cmd-F inside the document. The words become pixels. There's no undo button on a rasterized PDF.
How to use the PDF Compressor
Drop or pick your PDF. The tool accepts up to 100 MB and 200 pages — enough for almost every real-world document. Larger files run into browser memory limits; for those, use a desktop tool.
- Drop or pick your PDF. The tool reads the page count and file size and shows them before anything else.
- Read the warning. If the warning detects that your PDF is text-heavy, it'll say so. If it's mostly images, it'll tell you that too. The tool is biased toward not letting you wreck a document you cared about.
- Pick a preset. High quality (150 DPI, 85% JPG) is closest to the original. Balanced (100 DPI, 75% JPG) is what most people actually want. Aggressive (72 DPI, 60% JPG) is screen-only and will visibly degrade text edges.
- Click Compress PDF. Each page renders to a canvas at the chosen DPI, encodes as JPG at the chosen quality, and gets embedded in a fresh PDF. A progress bar shows page-by-page.
- Click Download. The result block shows before-size, after-size, and percentage reduction. The output is named
original-name-compressed.pdf.
Your PDF never leaves your browser. Pages render via pdfjs-dist, the JPGs come from the browser's native canvas encoder, and the new PDF is built by pdf-lib — all WebAssembly and JavaScript, no network round-trips during the compression step. Open your browser's network tab during the run and you'll see zero outbound requests.
The three presets, and which one to pick
The DPI controls how many pixels per inch the page gets rendered at. The JPG quality controls how aggressively those pixels get compressed afterward. Both knobs interact — lowering one and raising the other often produces similar file sizes with different failure modes.
| Preset | Render DPI | JPG quality | Typical size drop | Best for |
|---|---|---|---|---|
| High quality | 150 DPI | 85% | 30-50% | Print-ready output, archival, anything you'll look at closely |
| Balanced (default) | 100 DPI | 75% | 50-70% | Email attachments, sharing, viewing on screen |
| Aggressive | 72 DPI | 60% | 70-90% | Quick previews, fitting under a tight upload cap, never printing |
The simplest decision rule: if you'll print the PDF, pick High. If you'll share or read it on a screen, pick Balanced. If you have an upload limit you're about to fail and you can live with fuzzy text edges, pick Aggressive — and assume the result is single-use.
A worked example with real numbers
Take a real case: a 38-page scanned PDF of a signed contract, originally 47.2 MB. The pages were scanned at 300 DPI as full-quality JPGs — way more resolution than anyone needs for a screen-reading PDF. Here's what the three presets produce:
| Preset | Output size | Reduction | Quality notes |
|---|---|---|---|
| High (150 DPI, 85%) | 14.1 MB | 70% | Indistinguishable from original on a 14" laptop screen, prints cleanly |
| Balanced (100 DPI, 75%) | 5.8 MB | 88% | Slightly softer on close inspection, fine for screen reading and most printing |
| Aggressive (72 DPI, 60%) | 2.9 MB | 94% | Visible JPG artifacts on text, fine for a quick share, bad for printing |
The original was 47.2 MB — too big for most email systems (Gmail caps attachments at 25 MB, Outlook at 20 MB). After Balanced compression it's 5.8 MB and goes through anywhere. The signed contract still reads cleanly. The text selection that was in the original (added by OCR after scanning) is gone, but no one needed it for a contract you only have to forward.
Now the same tool on a different input: a 4 MB academic paper with real text and equations. After Balanced compression it's 6.2 MB — bigger. Rasterizing vector text is less efficient than keeping it as text, and the tool warns when output is larger than input. Same tool, wrong file type. That's why we put the warning up front.
How this compares to iLovePDF, SmallPDF, Adobe
The honest comparison: those tools do more sophisticated compression than we do, at the cost of uploading your file to their server.
iLovePDF and SmallPDF compress server-side using ghostscript or a similar PDF-native optimizer. They recompress embedded images while preserving text and fonts. Output is usually smaller than ours on text-heavy PDFs and similar or slightly larger on image-heavy PDFs. They also have free-tier limits (file count per day, file size cap) and the standard "Upload to remove limits" upsell. Your PDF lives on their servers for at least a few hours and possibly longer depending on retention policy.
Adobe's web compressor does the same thing iLovePDF does, in nicer packaging, behind a sign-in wall, with a quota that funnels you toward a Creative Cloud subscription. It's the cleanest output of the bunch on text-heavy PDFs. It's also the slowest, the most insistent about your email address, and the most expensive once you outgrow the free tier.
We're the opposite trade. Slower than nothing in the browser, but no upload, no account, no quota, no retention. We rasterize because that's what we can do well in-browser today; we tell you when that's the wrong choice. When we ship a real in-browser PDF optimizer that preserves text (it's on the list), it'll appear next to this one with the trade-off labeled.
What gets lost when a PDF is rasterized
Several things, and it's worth knowing all of them before clicking Compress:
- Text selection — you can no longer highlight, copy, or search words in the PDF. Every word is a group of pixels inside an image.
- Embedded fonts — fonts are no longer part of the file. The rendered pixels include the shapes of the letters, but not the font metadata. Re-flowing the text or changing the font is impossible.
- Hyperlinks — clickable links and bookmarks rasterize into the image. The blue underlined text is still visible but it doesn't go anywhere.
- Form fields — input fields, checkboxes, signature blocks all flatten into the page image. The PDF is now read-only.
- Vector graphics — charts, diagrams, and other vector elements turn into pixels. Zooming in shows pixelation instead of clean lines.
- Accessibility tags — screen readers can't read a rasterized PDF unless OCR is reapplied. This is a real cost for accessibility-critical documents.
If any of those matter for your document, this isn't the tool. Use our Split PDF tool to break the file into smaller pieces if you're trying to get under an upload limit, or use a desktop compressor like Adobe Acrobat that does proper PDF optimization.
Related PDF tools
The PDF Compressor is one tile in a larger PDF toolset. A few neighbors that often come up:
- Split PDF — break a large PDF into smaller files by page range. Often a better fix than compression when you only need to share part of a document.
- PDF to JPG — same rendering engine as the compressor, but it outputs the individual page images instead of repacking them into a PDF. Useful when you want the pages as standalone files.
- PDF Merger — combine multiple PDFs into one. Often paired with the compressor when assembling a multi-source document.
- Delete PDF Pages — drop unneeded pages before compressing. Faster than compressing, and it's the right first step if your PDF has 200 pages and you only need 10.
- Image Compressor — if what you really have is a folder of JPGs you were about to convert to PDF, compress them first and you'll skip the PDF-rasterization step entirely.
Frequently asked questions
Why does compression remove text selection?
Because this compressor works by rasterizing each page — rendering it as an image, then storing the image inside the PDF. Once a page is an image, there are no text objects anymore; the words are pixels. Most online PDF compressors that promise "preserve text" either don't actually compress much, or they require you to upload your PDF to their server where they can use more sophisticated tools. We chose: in-browser, honest about the trade-off, big size wins on the right input.
When is this a good tool to use?
Image-heavy PDFs are the sweet spot: scanned documents (receipts, contracts, signed forms), photo PDFs (vacation albums, real-estate listings), screenshot collections, anything with mostly raster content. These often compress 70-90% with no visible quality loss. The text selection you'd lose is usually fake-OCR layered on top of a scan anyway.
When is this a bad tool to use?
Text-only PDFs (books, contracts, articles, technical manuals) shouldn't be compressed this way. You get a smaller file but it becomes unsearchable, the text edges turn fuzzy under aggressive presets, and the reader can no longer copy quotes. For these, the right tool is either no compression at all (PDFs born from text are already small) or a server-side compressor that recompresses embedded fonts and images separately.
Is my PDF really not uploaded?
Correct. Pages render via pdfjs-dist in your browser, get encoded as JPG using the browser's native canvas, and the new PDF is built with pdf-lib — all in WebAssembly and JavaScript. Zero outbound requests during the compression step. Check your browser's network tab if you want to verify.
What size reduction can I expect?
Highly variable. Image-heavy PDFs at the Aggressive preset often drop 80-95% (a 50 MB scan becomes 3-8 MB). Text-heavy PDFs may actually grow because rasterizing vector text is less efficient than keeping it as text — the tool warns you if the output is larger. Most users land 50-70% smaller with the Balanced preset on the right type of input.
Does the output still print well?
Yes for the High and Balanced presets — 150 DPI and 100 DPI both print clean at standard office printer settings. The Aggressive preset (72 DPI) is screen-only; printing it will show visible pixelation, especially on text. Pick High if printing is part of the workflow.
Can I compress a password-protected PDF?
No — pdf-lib refuses to open encrypted PDFs. Remove the password first using a desktop reader (or whatever password-removal tool you trust), then compress the unprotected copy.
Will the page count or order change?
No. The compressor preserves the original page count and order exactly — every page in the input becomes one page in the output, rasterized in place. Bookmarks, hyperlinks, and form fields don't survive (they get rasterized into the page image), so this tool is for sharing and archiving, not for documents that need to stay interactive.