ToolsTray

Star a tool to keep it here.

Compress PDF

Compress a PDF by re-encoding the pictures inside it, down to a KB ceiling if a form demands one. Text stays sharp.

The document

Drop a PDF and its pages show here

Document

Drop PDFs here

One file or a whole batch — only the pictures are re-encoded, never the text — and nothing uploads

The level — each one names what it does

How much

Preset picks a level; Custom sets the quality and the picture size yourself; Target finds the largest file under a size.

Extras

Pictures lose their colour; the text keeps its own.

Clears the title, author, subject, keywords, producer and creator.

You save

—

How to use

  1. Drop your file onto the Document box. Its pages show beside the settings, with a rail to move between them. More than one is fine; they line up in a queue and run one after another.
  2. Choose How much. Balanced is preselected under Preset and covers most reports and scans; Target is the one to switch to when a portal has handed you a number in KB to beat.
  3. Press Compress PDF. The saving and the new size appear together, the page well shows the compressed file, ready to zoom into a photo and check it, Show the original flips back for comparison, and Download saves it as report-compressed.pdf.

About this tool

Weight in a PDF is nearly always pictures. So when you compress a PDF here, the pictures are the only thing touched: every embedded image is decoded, scaled down if it is bigger than the page needs, and re-encoded at the quality you asked for. Text and vector drawings are copied across as they stand. That is why the words in the file you get back are still selectable, and still sharp when you zoom to 400%.

How much is three choices: Preset, Custom and Target. Under Preset, Balanced is chosen for you when the page opens; it re-encodes photos at JPEG quality 62 and caps them at 1600 pixels on the longest side. Strong drops to quality 45 and 1000 pixels, Light climbs to 80 and 2200, and each card says so. If you would rather set both numbers yourself, Custom gives you a quality slider from 20 to 95 and a Max image side anywhere between 300 and 4000 pixels.

Target size is the option for a form with a hard cap. Type 100 into the KB box and five preset strengths get tried against your file, from quality 28 at 700 pixels up to 80 at 2200, keeping the best-looking one that still lands under the limit. If none of them fit, you still get the smallest version and the page says how far over the limit it landed. Feeding it a stack of PDFs works much the same: each file gets a row with its first page, its page count and, once it runs, its saving. A file that cannot shrink is kept as it was and said apart from one that failed, and Download all bundles the rest into compressed-pdfs.zip.

Three checkboxes sit under the choice. Convert images to grayscale earns its keep on scanned text, where the colour channels are mostly recording paper tint. Remove metadata empties the title, author, subject, keywords and the fields naming the software. Add page numbers, with a format and a corner to pick, numbers the document while it is open anyway, and the preview shows the numbers as you change them.

What it can't shrink

  • A report of text and tables has nothing to give up. When the rebuild comes out no smaller, the page says so, keeps your original, and offers Strong if you have not tried it.
  • Images carrying transparency are skipped on purpose, since re-encoding them as JPEG would flood the see-through parts with white. CMYK images are skipped as well, which is why press-ready files often come back barely changed.
  • Recompression is lossy on photographs. At Strong they are re-encoded at quality 45 and capped at 1000 pixels, which is visible on a screen-grab-heavy deck if you look for it. Text and line work are never re-encoded at all.
  • A PDF with an open password cannot be read, and the page says so as soon as you drop it. Take the password off with Unlock PDF, then compress what comes out of that.

Frequently asked questions

How do I compress a PDF to 100 KB?

Choose Target under How much and type 100 into the box that appears. Five preset strengths get worked through and the best-quality one that still fits under your number is what you download. Scans and photo decks usually get there. A long text report often won't, and then you get the smallest version with how far over it landed, which tells you whether to split the file instead.

Why is my PDF the same size after compressing?

Look at what is inside it. Ordinary JPEG photos are the easy case and always get rebuilt. Flate-compressed images get rebuilt too, but only where the renderer can hand over exactly one image at that pixel size. Anything with a transparency mask, anything in CMYK, and any replacement that would come out fatter than what is already in the file are all left alone. A document of text and vector charts has almost no image bytes to take out to begin with.

Will the text go blurry?

No. Text is never re-encoded, converted or rasterised; it is copied from the old document into the new one. Softness only ever turns up in photographs, and how much depends on the level you picked. The preview after a run is the place to check: zoom in on a paragraph, then on a photo.

Which setting is right for a scanned document?

Balanced, most of the time. A scan is a photograph of paper, so it compresses well and rarely needs to survive close inspection. Ticking Convert images to grayscale on top is worth a try when the pages are black ink on white, since the tint a scanner picks up off blank paper is noise you are paying to store. Keep Light for anything where the photo is the point, like a portfolio or a product sheet.

What does Remove metadata actually clear?

Title, author, subject, keywords, and the two fields naming the software the file came from. The hidden XMP block goes on every run whether the box is ticked or not. Dates are the exception, and the checkbox label oversells that part: the original creation date is carried over, and the modification date is stamped with the moment you compressed it. Treat the download as a new file rather than an anonymous one.

Can I compress a folder of PDFs in one go?

Select as many as you like and they run in sequence with a progress bar and a Cancel button that keeps whatever already finished. Each lands in its own row showing what it started at and what it came down to. A file that cannot shrink is marked skipped and kept as it was, one that cannot be opened is marked failed, the queue carries on, and Download all wraps the rest in compressed-pdfs.zip.

Missing something in Compress PDF? Suggest a feature →

Tell someone who needs this.