PDF: free online tools
9 free tools, processed locally in your browser.
Nine tools share one file format here, but not one internal mechanism, and that detail decides which of them you want. Merge, Split, Rotate and Sign open the document with pdf-lib and copy the pages across as they are: the text stays selectable, nothing is re-encoded. Compress PDF in Scan mode, Protect PDF and PDF → JPG take the other route, through PDF.js: each page is painted into a canvas and the output file is rebuilt out of those images. PDF → Word and the OCR tool produce no PDF at all — they pull text out of one.
Knowing which side a tool sits on avoids the classic disappointment, a document that comes back with text nobody can copy any more. What follows sorts out the tools that look alike from a distance, and shows the order to run them in when a job needs more than one.
Tools in the PDF category
- Compress PDF — Shrink a PDF with an adjustable compression level.
- Merge PDF — Combine several PDFs into one document, in your order.
- Split PDF — Extract pages or cut a PDF into several files.
- PDF → Word — Convert PDF text into an editable .docx document.
- PDF → JPG — Export each PDF page as a high-quality image.
- Protect PDF — Lock your document with a password.
- Sign PDF — Draw your signature and place it on any page.
- OCR — scanned PDF — Extract text from a scan to copy and search it.
- Rotate PDF — Fix the orientation of some or all pages.
Which tool for which job
A file too heavy for an email attachment is a job for Compress PDF. Several documents to bring together: Merge PDF, which follows the order shown in the list, adjustable with the arrows before you start. Three pages to isolate out of sixty, or a batch scan to break into separate files: Split PDF and its two modes. Pages lying on their side or upside down: Rotate PDF, which accepts a page range and leaves the rest untouched.
Reusing the content depends entirely on how the document was made. A PDF born in a word processor carries a real text layer, which PDF → Word reads back to rebuild a .docx. A PDF that came out of a scanner holds nothing but images, so OCR is the only way in. Sign PDF, finally, places a signature you draw by hand, cropped to the stroke itself, at one of six offered positions.
Compress, or convert to images?
Both of these tools make a PDF lighter, but not at the same price. Compress PDF runs Ghostscript compiled to WebAssembly for the Maximum, Balanced and High quality levels: it resamples the images inside the file and leaves the text and vector artwork in place, so the result is still readable and still indexable. PDF → JPG renders every page as a separate image, which is convenient for illustrating a slide deck and unsuitable the moment somebody has to read or search the document.
The Scan level of the compressor occupies the middle ground: it rasterises the pages, which flattens everything, text included. Keep it for documents that were already scans. One safeguard is worth knowing about: if the compressed version comes out heavier than the original, the tool hands you the original back and says so. A PDF made of nothing but text has nothing left to give up.
Chaining several tools without trapping yourself
Order matters. On a crooked scan, rotate first and run OCR second: Tesseract reads tilted text badly. OCR returns a .txt file, not a searchable PDF, so if the point was to make the scan itself lighter, send it through the compressor in Scan mode afterwards. With a case file drawn from several sources, merge before compressing rather than the other way round.
The signature and the password always come last. Protect PDF rebuilds the document as images before encrypting it, and printing is the only permission left open. The pdf-lib tools refuse to open a protected file at all, so encryption really is the end of the chain.
Useful vocabulary and common mistakes
Two notions settle most cases. A text PDF carries real characters, an image PDF only pixels: the first converts to Word, answers a keyboard search and compresses well, while the second needs OCR first. When PDF → Word reports that no text was detected, the diagnosis is made. The second notion is resolution in dpi: the compression levels aim at roughly 96, 150 and 300 dpi, the last being the usual floor for clean printing.
On the input side, the range fields in Split, Rotate and OCR share a single syntax: page numbers and intervals separated by commas, of the 1-3, 5, 8-10 kind. Since page-by-page work happens in memory, a document of several hundred pages is easier to handle if you split it before processing it.