Re-encode photos and scans inside a PDF on your device. Text-heavy files often shrink less than image-heavy ones.
This tool shrinks photos and scans inside the PDF. Text-heavy documents often change little because the letters are not images.
PDF Compress makes a PDF smaller by re-encoding the raster images embedded in it — photos and many scans — as JPEGs, and by shrinking images whose long side is above the limit you set. It does not take a screenshot of each page. Text, vector lines, and fonts stay as they are, so a sentence you can select should still be selectable afterward. That is why a text-heavy contract often barely changes size, while a phone scan of a receipt can drop a lot.
The tool walks the PDF’s internal objects with pdf-lib, finds image streams, draws the ones it understands onto a canvas, and writes a JPEG back only when the new bytes are smaller than about 97 percent of the original image stream. If the new JPEG is not clearly smaller, the original image is left untouched. The saved file uses object streams, and the producer and creator fields are set to QuickConvert. The download name is your file name with the extension compressed.pdf, so scan.pdf becomes scan.compressed.pdf.
Three presets fill in the two controls. Strong uses JPEG quality 42 and a 1000 pixel long edge. Balanced, the default, uses quality 62 and 1500 pixels. Light uses quality 80 and 2200 pixels. You can still move the quality slider between 30 and 90 and set the max edge between 400 and 4000 pixels. Images that already fit inside the max edge are not upscaled. Images that have a soft mask are re-encoded but not downscaled, so a transparent cutout keeps its pixel size.
Some images are skipped on purpose: masks, images smaller than 32 pixels on a side or under 8,000 pixels in area, images that are not 8 bits per component, JPEG 2000, JBIG2, CCITT fax, images with a hard mask, and non-JPEG images whose color space is not DeviceGray or DeviceRGB. A skip is not a failure. The summary after a run tells you how many images were found and how many were re-encoded.
Each file must be 80 MB or smaller. The page explains, in the summary, when no embedded photos were found.
Run Balanced first and open the download next to the original. Look at small text that was part of a photo, such as a photographed document, because that text is pixels and JPEG will soften it. Text that was real text in the PDF is not put through JPEG.
If the summary says no embedded photos were found, another pass with Strong will not invent savings. The file is mostly text, vectors, or an image encoding this tool skips. Exporting from the original application, or reducing images before they were placed, is the next step — this page cannot rewrite a font into a smaller font.
Scans of color photos shrink more than black-and-white fax-style CCITT images, which are skipped. A mixed file may shrink only in part, and the summary’s “re-encoded N of M” line is the honest account of what changed.
For a loose photo that is not inside a PDF yet, Image Compress is the direct tool. Compress the pictures, then place them, if you still control the layout.
The size stayed the same. Either there were no compressible images, they were already small, the new JPEG failed the “worth replacing” check, or the encodings were skipped. The summary distinguishes “no embedded photos” from “processed some images but the file size stayed about the same.”
The file grew a little. Saving a PDF can reorganize objects. The tool only replaces an image when the replacement is smaller, but the container can still differ by a small amount. If the summary shows no real saving, keep your original.
Password-protected files are refused. Unlock them in a reader first. QuickConvert never asks you to type the password into this site.
A damaged PDF, or a file that only pretends to be a PDF, fails at load with a message that it could not be read. Unusual images that throw while decoding are skipped rather than failing the whole file, when the rest of the document can still be saved.
The browser tab uses a lot of memory on a large scan because each image is decoded on the canvas. That work is local. If it fails, try a shorter file or a computer with more memory.
Compression runs entirely on your device. The PDF is parsed in the browser, images are drawn with the canvas, and the new file is downloaded from memory. QuickConvert’s servers do not receive the document, and there is no queue you wait in. The temporary download URL is revoked shortly after the save starts. Close the tab and the working copy is discarded.
Analytics and ads on the public site can still see that the page was opened. They do not see the PDF. Read the Privacy Policy and how to reduce a PDF’s size for the quality tradeoffs in more detail.
A PDF can contain text, fonts, vector drawings, and pictures. On most large PDFs, the pictures are what take the space: a phone scan is a set of full-page photos. PDF Compress finds those embedded images, redraws them as JPEGs at the quality you choose, and shrinks any that are larger than the max image edge. It replaces an image only when the new one is clearly smaller. Text, fonts, and vectors are not changed, so selectable text stays selectable. Everything happens in your browser; the PDF is not uploaded.
The max image edge is the long side, in pixels, that an embedded image is allowed to keep. On an A4 or US Letter page, that number translates roughly into the sharpness you will see:
You can also set quality from 30 to 90 and the edge from 400 to 4000 pixels yourself. A useful middle setting for strict upload limits is quality 55 with a 1200 pixel edge.
University, visa, and job portals often accept 2 MB or 5 MB per file. Divide the cap by the page count to get a per-page budget, then work down from Balanced. The guide to getting a PDF under an upload limit shows the full method.
Many mail services cap a message at around 20 to 25 MB, and a long colour scan can exceed that. Balanced usually brings a phone-scanned document well under it.
For receipts and letters you keep for records, Balanced or Strong saves storage while keeping the content readable. Keep originals of anything you might need to print at high quality.
A brochure or report that embeds full-resolution camera photos can shrink a lot, because those images are far bigger than the page needs.
Usually not. This tool saves space by compressing images inside the file. Letters, vectors, and fonts are left alone, so a text-heavy PDF may barely change.
What actually makes a PDF large, which QuickConvert preset to start with, and why a text-only file often will not shrink.
How to tell a browser-side tool from an upload, and what QuickConvert still sends when your files stay on the device.
A practical way to shrink photos before you attach them, including what the quality slider and long-edge cap actually do.
A step-by-step way to fit a PDF under a fixed upload cap: work out a per-page budget, pick the right PDF Compress preset, and know when to split or rescan instead.