Compress PDF Experimental
A browser-only attempt at reducing PDF file size by re-encoding each page as a compressed JPEG image. It can help with scanned and photo-heavy documents. It removes selectable text and it cannot shrink every file, so this page reports the real byte sizes and tells you honestly when the result is bigger.
Your files are processed locally in your browser tab. They are not uploaded to this server and they are not sent to any third party.
Read this before you run the tool. Compression here works by turning each page into an image. Your text stops being text, so it can no longer be selected, copied, searched or read by assistive technology. Links, bookmarks, form fields, annotations and digital signatures are not carried over. For a document that is mostly text the result will often be larger than the original. Keep your original file and check the downloaded copy before you rely on it.
- File
- Size and pages
Compression level
Measured result
- Original file
- —
- New file
- —
- Difference
- —
How this compression actually works
- Add the PDF file you want to try to reduce.
- Choose a compression level and read what it does to the pages.
- Tick the confirmation checkbox to acknowledge what will be lost.
- Press the button and wait while every page is rendered one at a time.
- Compare the two file sizes shown in the result. Keep whichever file is better for you.
Every page is drawn onto a canvas at the chosen scale, encoded as a JPEG at the chosen quality, and the images are placed into a brand new PDF with one page per source page. No attempt is made to recompress the original embedded images in place, to subset fonts, or to remove unused objects. That is the honest limit of what can be done reliably without a server-side optimiser or a paid conversion API.
When this helps and when it does not
| Document type | Typical outcome | Why |
|---|---|---|
| Scanned pages, photographed receipts, image-heavy brochures | Usually smaller | The pages are already pictures, so re-encoding them at a lower resolution removes real data. |
| Plain text reports, contracts, invoices generated as text | Usually larger | A few kilobytes of text is replaced by a full-page image of the same page. |
| Documents exported from Word, LaTeX or InDesign with embedded fonts | Usually larger | Fonts and vector art are already efficient; rasterising them is much heavier. |
| Files already saved with strong compression | Unpredictable, often larger | There is little redundancy left to remove, and JPEG adds its own overhead. |
The table describes typical behaviour, not a guarantee. The result panel shows the measured sizes for your own file, and that is the only figure this site relies on.
What this tool does not do
- It does not preserve selectable, searchable or accessible text.
- It does not keep links, bookmarks, outlines, annotations, form fields or digital signatures.
- It does not guarantee a smaller file, and it says so explicitly when the result grows.
- It does not use a server-side optimiser, so it cannot recompress embedded images in place.
- It does not process encrypted PDFs and it does not remove passwords.
Frequently asked questions
Why is this tool marked experimental?
Because the only technique available in a browser without a paid server service is destructive: each page is turned into a JPEG image and written into a new PDF. That can shrink a scanned or photo-heavy document a lot, and it will usually make a text-heavy or already-optimised document larger. Publishing it as a normal "compress PDF" button would be misleading, so it is labelled and explained instead.
What exactly is lost when I use it?
Selectable and searchable text, because words become pixels. Vector graphics are flattened, so lines and charts can look slightly soft. Links, bookmarks, form fields, annotations, digital signatures and document metadata are not carried over. Embedded fonts are gone because there is no longer any text to draw.
Will it always make my file smaller?
No, and the tool never pretends otherwise. It measures the real byte size of your original file and of the new file and shows both numbers side by side. If the result is larger, it says so and recommends keeping your original.
Which documents benefit most?
Documents made of scanned pages or photographs, where the pages contain very little text: scanned contracts, photographed receipts, image-heavy brochures. Documents that are mostly text, or that were already saved with good compression, normally get bigger.
Can I still select the text afterwards?
No. After the conversion the text is part of an image. The page cannot be searched, copied from, or read by a screen reader. If you need searchable text, do not use this tool; keep the original file instead.
How does this differ from what a desktop PDF editor does?
A desktop tool can recompress just the embedded images, drop unused objects, subset fonts and rebuild the cross-reference table while leaving text as text. Doing that reliably in the browser would need a much larger PDF manipulation library than this site ships, so this tool does not claim to match it.
Is my document uploaded for this?
No. The pages are rendered by your browser in the same way as the PDF to JPG tool. Nothing is sent to the server. That is also why the tool cannot use a more sophisticated server-side optimiser.