Skip to content
user@convertitive:~$convertitive

Compress PDF

Drop a scanned PDF, pick a quality, download a much smaller copy. Nothing leaves your browser.

This tool compresses a PDF the blunt-but-effective way: it renders each page to pixels with Mozilla’s pdf.js, re-encodes that rendering as a JPEG at a quality you control, and rebuilds a fresh PDF around those JPEGs. On scanned documents and photo-heavy decks this routinely cuts 70–90% of the file size. But be clear about the trade-off: the output is a flattened copy. Text becomes an image — no longer selectable, searchable, or copy-pasteable — and any form fields, links, or bookmarks are discarded. The result opens everywhere and prints fine, but keep your original for archival. On born-digital, text-only PDFs this approach can actually make the file bigger; the tool will tell you honestly when it does.

Drop a PDF here, or

One PDF, up to 64 MB and 100 pages. Stays in your browser — nothing is uploaded.

How to use

  1. Drop your PDF

    Drag a single PDF (up to 64 MB and 100 pages) onto the dropzone, or click to choose a file. It is opened locally with pdf.js — nothing is uploaded.

  2. Pick a JPEG quality

    Lower quality means a smaller file. 50–60% is usually plenty for scanned documents; go to 70% or higher if the pages contain photos you care about.

  3. Choose a Detail level

    Detail controls the rendering resolution. 1× gives the smallest file, 1.5× is a good balance for reading on screen, 2× keeps small print sharp at the cost of a larger file.

  4. Compress and download

    Click Compress and watch the per-page progress. When it finishes you see the exact before/after size and can download the result as a new PDF.

Frequently asked questions

Does my PDF get uploaded to a server?
No. Every step — parsing the PDF, rendering pages, JPEG encoding, and rebuilding the output file — runs in your browser with JavaScript. There is no upload request at all, so Convertitive never sees your document.
When does this compression work well, and when does it work badly?
It works well on scanned documents, photographed pages, and image-heavy slide decks — anything where the pages are already essentially pictures. There, JPEG re-encoding routinely saves 70–90%. It works badly on born-digital, text-only PDFs: their text is stored as compact vector instructions measured in kilobytes, while a rasterized page is millions of pixels measured in megabytes. Rasterizing such a file usually makes it larger, and the tool will say so instead of hiding it.
Does this remove password protection?
No. Password-protected (encrypted) PDFs are rejected with an error — this tool cannot and does not bypass encryption. Remove the password with the tool that applied it first, then compress the unprotected copy.
What quality setting should I use?
For black-and-white or plain text scans, 50–60% is nearly indistinguishable from the original at reading size and gives the biggest savings. If the pages contain photographs, charts with subtle gradients, or anything you might zoom into, use 70% or higher. Below about 40% you will start to see visible JPEG blockiness around text edges.
Is the text (or OCR layer) preserved?
No — this is the most important trade-off to understand. The output pages are images, so the original text layer, including any OCR layer added to a scan, is discarded. You can no longer select, search, or copy text in the compressed file. If searchability matters to you, keep the original alongside the compressed copy, or use a tool that compresses images inside the PDF without rasterizing the text.
What are the file and page limits?
One PDF at a time, up to 64 MB and 100 pages. Because everything runs in your browser's memory, very large or very long documents would otherwise risk crashing the tab. Note also that a higher Detail setting produces pages with physically larger nominal dimensions (the page is sized to the rendered pixels at 96 DPI) — that is harmless for on-screen reading, and printing scales to paper automatically.

About

Why text PDFs don't compress this way

A born-digital PDF stores text as font references and positioning operators — a full page of prose might occupy two or three kilobytes, plus a shared, already-compressed font file. That representation is resolution-independent and close to optimal. Rasterizing it replaces those few kilobytes with a grid of pixels: at 1.5× render scale a Letter page is roughly 1,200 by 1,600 pixels, and even a well-compressed JPEG of it lands somewhere between 100 and 400 KB. Multiply by every page and the 'compressed' file dwarfs the original. That is not a flaw in the implementation — it is the fundamental economics of vectors versus pixels, and it is why this tool reports honestly when the output grows instead of shrinking.

What professional tools do differently — and when flattening is actually what you want

Tools like Ghostscript or Acrobat's own optimizer compress selectively inside the PDF structure: they find the embedded image objects, downsample and re-encode just those, subset fonts, and strip unused objects — all while leaving text as text. That preserves selectability and searchability and is the right approach for mixed documents. But full rasterization has legitimate uses beyond size: a flattened copy removes hidden layers, cropped-out image regions, embedded metadata, form field logic, and revision leftovers — everything except what is literally visible on the page. If you are sharing a document and want certainty that nothing invisible travels with it, a flattened copy is a feature, not a compromise.

Sources & references

Authoritative references behind the math, constants, and tables on this page. Verified by Buğra Sözeri on the dates shown and re-checked at every deploy.