Your images are processed in your browser — Cloud HD (Pro, opt-in) is the only exception

Compress a scanned PDF

Drop a PDF, pick how hard to squeeze, get a smaller one back. Pages keep their exact size. Best on scans, because every page comes back as an image. Nothing is uploaded.

📄
Drop a PDF here
or click to browse · .pdf up to 200 MB

What comes back. One PDF, same number of pages, each page the exact size it was — an A4 page returns 595.28 × 841.89 pt, not a rounded approximation of it. The content of each page is a JPEG photograph of the original page rendered at twice its nominal size, so it stays sharp on screen and at normal print sizes. Selectable text does not survive; it becomes part of that picture.

Limits, honestly. Files up to 200 MB are accepted, and that cap is enforced at the point you choose the file. The limit you are more likely to meet first is page count rather than megabytes: every page is held in memory at once while the new document is written, so a several-hundred-page scan can exhaust a phone before it exhausts the size cap. If a long document stalls, split it and compress the halves.

If it cannot win, it says so. The finished document is measured against the one you gave it. If the result is not smaller, you are told both numbers and given nothing to download — keeping your original is the right answer, and handing you a bigger file labelled “compressed” would not be.

Compress PDFFree, Fast & Private

This compresses a PDF the way a PDF compressor built on a browser can: it renders every page, re-encodes each one as a JPEG at the quality you chose, and rebuilds the document around them. One file in, one file out, nothing uploaded. A 10 MB scan of a certificate set typically comes back at 1-3 MB.

The trade-off is real and worth stating before you start. Rebuilding this way rasterises the whole document, so any selectable text becomes part of the page image. Text stops being searchable and copyable, and the file will not pass a portal that requires machine-readable text. For a scanned document that was already an image, you lose nothing. For an exported report or a contract with live text, do not use this route — export it again at a lower image quality from whatever produced it. The tool checks the first few pages for live text when you choose the file and warns you before you commit, rather than after.

Page geometry survives exactly. An A4 page goes in at 595.28 × 841.89 pt and comes back at 595.28 × 841.89 pt, not a rounded approximation and not the 2.7×-inflated page this round trip used to produce before v4.33.0. That matters if you are going to print the result, or submit it somewhere that validates A4. Resolution is what you are trading away; page geometry is not.

It can also lose, and it says so when it does. Photographing a page of sharp text costs more bytes than the instructions for drawing that text did, so a born-digital document can come back bigger than it went in. When the finished file is measured and turns out not to be smaller, you are shown both numbers and given nothing to download. Keeping your original is the correct outcome in that case, and handing you a larger file under a Compress heading would not be.

Everything runs in your browser using PDF.js to render and jsPDF to rebuild. Your documents never leave your device, which is the point for legal contracts, financial statements, medical records, and ID paperwork.

Most email services cap attachments at 25 MB, and many form portals want PDFs under 5 MB or 2 MB. If you need to land under a specific kilobyte number rather than just smaller, the size-targeted PDF builder aims at a byte budget instead.

How it works

  1. Drop your PDF: Drag a PDF onto the box above, or click to browse. Files up to 200 MB are accepted, and nothing is uploaded — the file is read in your browser.
  2. Read the warning, if there is one: MiniPx checks the first pages for selectable text. If it finds any, it tells you before you run: this method turns that text into pixels permanently, and usually makes such files bigger rather than smaller.
  3. Pick how hard to compress: Balanced keeps small print readable. Smaller softens fine print and suits most upload portals. Smallest is for screen reading only.
  4. Compress and check the numbers: Every page is rendered, re-encoded and rebuilt at its original page size. You get the before and after byte counts. If the result is not smaller, nothing is offered for download and you keep your original.

Frequently asked questions

Does MiniPx have a one-click PDF compressor?
Yes, on this page, since v4.34.0. Drop a PDF, choose a compression level, and get one back. Before that this page said it did not, and its own FAQ said it accepted 200 MB files — the tool now exists, so both statements can finally be true at once.
Will my text stay selectable?
No. Rebuilding from page images rasterises the document, so text becomes part of the image. This is permanent and one-way. MiniPx checks the first pages when you select the file and warns you if it finds live text, so you can back out before running. If you need searchable or copyable text, or a portal demands machine-readable text, re-export from the original source instead.
Does the page size change?
No. Each page comes back at the exact size it went in as, down to the fraction of a point — an A4 page returns 595.28 by 841.89 pt. Earlier versions of this round trip inflated pages by about 2.7 times in each direction, so an A4 scan came back roughly A1. That was fixed in v4.33.0 and this tool is built on the fix.
What if the result is bigger than the original?
You are told, and you are given nothing to download. The finished document is measured against the file you supplied, and if it is not smaller the tool says so with both byte counts and tells you to keep your original. A rasterising compressor genuinely can produce a larger file on a text-heavy document, and handing that back under a Download button would be worse than refusing.
What is this actually good for?
Scanned documents. A scan is already a full-page image, so rasterising costs you nothing and the size drop is large — 50-70% is typical. Certificates, mark sheets, signed forms, and photographed paperwork are the sweet spot. Born-digital exports are the case where it can lose.
Is it safe for legal and financial documents?
Yes. Every step runs in your browser through PDF.js and jsPDF. Your documents are never uploaded to any server, so there is nothing on our side to leak or subpoena. Your original file is never modified either; the tool only reads it.
What is the maximum PDF file size?
Files up to 200 MB are accepted, and that cap is enforced when you choose the file rather than being an estimate. In practice page count binds sooner than megabytes: every page has to be held in memory at once while the new document is written, so a several-hundred-page scan can exhaust a phone well before it reaches the size cap. If a long document stalls, split it and compress the halves.
Can I compress multiple PDFs at once?
One PDF at a time. Each run reads a single document and produces a single document, so the page count and the page sizes of the result always match the file you gave it.

All Image Tools

AI Tools

Solutions by use case