Shrink scanned PDFs. Text becomes an image. Runs on your device.
This tool draws every page of your PDF as a picture, saves each picture as a JPEG, and builds a new PDF from those pictures. Each new page has the same printed size as the old one, so an A4 page still prints as A4. The price is that the page is now only an image. Text can no longer be selected, copied or searched, links stop working, form fields are removed, and bookmarks are dropped. The tool does not read the text and does not try to keep it. The three strengths set how the pictures are made. Light draws at 150 DPI with JPEG quality 75, Balanced at 120 DPI with quality 60, and Strong at 96 DPI with quality 45. An A4 page at 120 DPI is about 992 by 1403 pixels. You can also turn the pictures to greyscale, which removes colour and often makes the file smaller.
The best candidates are PDFs that already are pictures: a phone scan of a contract, a stack of receipts saved as a PDF, or photos placed on pages. Those files are large because each page holds a high-resolution image, and drawing them again at a lower resolution can cut the size a lot. How much depends on the pictures, so the tool shows the real before and after sizes when it finishes instead of promising a number. A PDF made from a word processor or a spreadsheet is different. Its pages are mostly text and vector shapes, which take very little space, and a picture of a page is usually bigger than the text it replaces. For those files the result can be larger than the original, and the tool then tells you so and does not offer the larger file as the result. A separate button still downloads the picture-only copy if you want it anyway.
Do not use this tool on a PDF you need to search, quote from, fill in, or click through. If the PDF is mostly real text, the tool shows a stronger warning before you start, because it replaces that text with slightly blurry pictures and the text is lost. It does not compress the original file in place, and it does not do lossless tricks such as merging duplicate objects or trimming fonts. It handles one PDF at a time and cannot open a password-protected file. If you only need some pages, split the PDF first. If you need the words from a scan as text, use an OCR tool instead, because this one does not read text at all.
Choose Target size, pick 1, 2, 5 or 10 MB or type your own limit between 50 KB and 50 MB, and the tool tries to compress a PDF to a size limit. It first draws up to three sample pages at each strength to guess how big the whole file would be, then starts at the gentlest strength that looks like it will fit: Light, Balanced, Strong, or Minimum quality at 72 DPI. It builds the real file and measures it. If that file is still over your limit, it steps down one strength and builds again, up to four builds in all, and stops at the first file that fits. The guess can be wrong, which is why the real file is always measured. Your result says which strength was used. If even Minimum quality is over the limit, the tool shows the smallest size it reached and offers that file only if it is smaller than your original, marked as over the limit. If your PDF is already under the limit, nothing is rebuilt. Very small limits on a scan make the pages blurry and hard to read, and a PDF made from a word processor usually cannot be reduced this way at all, because its pages are already small text.
The PDF is opened and drawn by code running in this tab. There is no upload step and no server that receives your pages. The page counts visits with Google Analytics, using rough groups such as 1 to 10 MB, never a file name, a page or an exact size. Pages are processed one at a time, and each page picture is released from memory as soon as it has been saved as a JPEG, but the finished JPEGs add up until the new PDF is built. A PDF over 200 MB or over 1000 pages asks you to confirm before it loads, and on a phone a PDF over 100 pages shows a warning because the device may run out of memory. If you cancel, nothing is saved, because the new file is only complete when every page is done.
Every page is turned into a JPEG picture and the PDF is rebuilt from those pictures, so the letters are now pixels, not text. Copying, searching and screen readers stop working on the new file. Keep your original if you may need the text, or run OCR on the result.
A PDF made from a document or a spreadsheet stores text and shapes compactly, and a JPEG of each page can be bigger than that. When the new file would be the same size or larger, the tool says the PDF is already compact and does not offer it as the result, so you never get a bigger file by mistake. If you want the picture-only copy anyway, a separate button downloads it; every page in it is a picture, so text can't be selected or searched.
Light draws pages at 150 DPI with JPEG quality 75 and keeps the most detail. Balanced uses 120 DPI and quality 60. Strong uses 96 DPI and quality 45, which gives the smallest file and softer small print. If small print is hard to read after Balanced, go back to Light.
Switch to Target size, choose 2 MB, and press Compress PDF. The tool estimates from a few sample pages, builds the file, measures it and steps to a stronger setting if it is still over, with at most four builds. You see which strength was used. If 2 MB cannot be reached, it tells you the smallest size it got instead of pretending.
Every page is a JPEG picture, and below a certain size the pictures become too blurry to read, so the tool stops at Minimum quality (72 DPI). A PDF that is mostly real text is usually already small, and a picture of its pages is often not smaller, so it cannot be reduced this way. Try splitting the PDF into parts, or keep the original.
No. The new pages are flat pictures, so clickable links, fillable fields, comments and bookmarks are all gone. If a form must stay fillable, do not compress it with this tool.
No. The PDF is opened and rebuilt in your browser tab, so no page leaves your device. The site counts visits with Google Analytics using rough size groups, never a file name or a page.
Keep the original, because turning text into pictures usually does not help. If the file is large because it has a few big pages, split those pages out first and compress only them. The tool warns you when it finds text on more than half of the pages it checks.