What is a Compress PDF?
Most large PDFs are large for one reason: photographs. A fifty-page report of text is a couple of megabytes; the same report with twenty full-page photographs is thirty. Text in a PDF is stored as font outlines and drawing instructions, so it is compact, and vector art is compact too. Images inside a PDF are usually stored as JPEG or Flate-compressed bitmaps at whatever resolution they were pasted in at — and pasted-in images are often 300 DPI photographs that will never be printed at more than screen size.
Compression therefore means, almost always, downsampling those embedded images: reducing a 300 DPI photograph to 150 DPI and re-encoding it at a lower quality. The visible difference on screen is usually nothing at all, and the file often shrinks by half or more. The text and the line art are left as they are, so a compressed contract stays as crisp and as selectable as it was.
There are three sensible degrees. Lossless compression rebuilds the file's internal structure and removes duplicated objects without re-encoding a single image — safe and reversible in spirit, but often a small win. Balanced compression re-encodes the photographs at a lower resolution and quality, which is where the 30–70% reductions come from. Maximum compression pushes harder and will rasterise a whole page if that is the only way to shrink it, which is what makes text stop being selectable.
How to reduce a PDF file size
- Choose the PDF. Its size is shown so you know what you are starting from.
- Pick a preset. Start with balanced: it is the best trade of size against quality for most documents, and the one that produces the dramatic reductions.
- Press Compress PDF and watch the progress. The file is processed in chunks, so long documents show steady movement rather than one long wait.
- Read the before and after figures. If the saving is small or negative, the document was already optimised or is mostly vector art.
- Download the compressed file. Keep the original until you have checked a page or two at full size, in case an image crossed the line from sharp to soft.
Compress PDF limits and rules
| Item | Limit |
|---|---|
| Presets | Lossless, balanced and maximum |
| Lossless saving | Typically 5–30%, occasionally nothing |
| Balanced saving | Typically 30–70%, by re-encoding the embedded photographs |
| Maximum saving | Typically 60–90%, and it may rasterise pages |
| Text and vectors | Left as vectors under the first two presets, so they stay sharp and selectable |
| Fallback behaviour | If a preset fails part-way the compressor retries with a lighter one and reports it |
| Chunk size | Ten pages at a time by default, so memory use stays flat |
| Timeout | Five minutes before the run is abandoned |
| Encrypted PDFs | Not supported; remove the password by printing to a new PDF first |
| Uploading | None — the compressor is WebAssembly running in your tab |
When to use it
- Getting a scanned document under an email attachment limit without splitting it into pieces.
- Reducing a portfolio or brochure PDF before putting it on a website, where a 30 MB download loses readers.
- Shrinking a file that must be uploaded to a portal with a hard 10 MB cap.
- Cutting the size of a photo-heavy report to send over a slow connection.
- Trying the lossless pass first on a document that must stay pixel-perfect.
Is it private?
The PDF is never uploaded. Compression happens with a WebAssembly build of Ghostscript running inside this tab, reading the file from local memory, so the document stays where it is. The one network request involved is downloading the compressor itself — a few megabytes of WebAssembly plus its loader from a public CDN — and that request contains none of your data. After the first use the compressor is cached by the browser, and everything works offline from then on.
Frequently asked questions
It depends entirely on what is inside it. A photograph-heavy brochure typically halves at the balanced setting and can shrink by 80% at maximum. A text-only document that was exported correctly in the first place may barely move, and a file that is already optimised can even grow slightly — re-encoding is not free. The result panel always shows the real before and after, so there is nothing to guess at.
Under the lossless and balanced presets, no. Text is stored as font outlines rather than pixels, and neither preset touches it — the pages stay sharp and the text stays selectable and searchable. The maximum preset is the exception: it can decide to rasterise an entire page, which makes the text part of the picture, so it stops being selectable and slightly softens. Use maximum only when the size target really is that strict.
No. The compressor is a WebAssembly build of Ghostscript downloaded into your browser, and it reads the file from local memory. The only network traffic is fetching the compressor itself, which is the same binary for every visitor. This is why the tool keeps working with the network disconnected once it has loaded.
Because the original was already well optimised, or because the document is mostly vector art and text that re-encoding cannot help — so the only effect of the pass is a slightly different internal structure. Nothing is wrong, and the tool tells you when this happens. Download the original.
Not directly. An encrypted PDF has to be decrypted before its contents can be read and re-encoded, and this tool deliberately does not ask for passwords. Open the file in a PDF reader where you know the password, print it to a new PDF, and compress that — the output is unencrypted and the pages are identical.
The lossless and balanced presets rebuild the document structure rather than the page content, so links, form fields and bookmarks generally survive. Rasterising under the maximum preset does not: a page turned into an image loses its interactive elements along with its selectable text. If the document needs to stay fillable, stay away from maximum.
Related tools
Written and maintained by Md. Nur Hossain Repon.