Reduce PDF file size by optimizing structure, with optional recompression of the JPEG photos inside.
If this tool isn’t working as expected, please take a screenshot of the error and report the problem here so we can investigate and improve it.
Large PDF files cause real problems: email attachment limits reject them, upload forms refuse them, and slow connections make sharing painful. The lossless pass is a safe first step - it carries no quality risk - and often shrinks PDFs exported from office software that carry a lot of metadata and structure.
For scans and photo-heavy PDFs, where the size is mostly image data, the image options recompress the JPEG photos so the file can shrink substantially. Either way, the tool checks that the new PDF opens again with the same number of pages before offering it, and tells you honestly when there is nothing to save.
This PDF compressor reduces file size in two honest ways. By default it does a lossless structure optimization: it strips embedded metadata (title, author, subject, keywords, producer, creator) and rewrites the file's internal structure with object streams, so visual quality is completely unchanged. Optionally, Light, Balanced or Strong also recompress the JPEG photos inside the PDF, which is where most of the size of a scanned or photo-heavy PDF lives. It is not a full PDF compressor like Ghostscript: it never rasterizes pages, and if the result isn't smaller you keep your original.
Choose the PDF file you want to compress.
Optimize the PDF structure, or also recompress its JPEG photos with Light, Balanced or Strong.
Download the result when it is smaller; if not, your original is kept.
Upload your PDF to see its current file size displayed immediately.
Leave Structure only for a lossless pass, or choose Light, Balanced or Strong to also recompress the JPEG photos inside the PDF.
Click Optimize PDF. With an image option you see the progress through the images, and afterwards how many were recompressed and why others were left alone.
If the file didn't get smaller, the tool shows "Original kept — no size reduction." and offers no download. Sizes and the percentage are calculated from the exact byte counts.
When the result is smaller, click Download Optimized PDF - the filename gets an "_optimized" suffix. Use Upload another PDF or Reset to start over.
Best For
Input
A PDF exported from word-processing software with embedded author/title/producer metadata and many internal objects, Structure onlyResult
A noticeably smaller PDF with the same pages, text and images, and the metadata removedStripping metadata and consolidating internal objects into compressed streams has real content to work with in this kind of file.
Input
A PDF made up of JPEG scans of pages, Balanced image recompressionResult
"Images recompressed and structure optimized" with a note such as "6 of 6 images recompressed", and a much smaller fileOnly the JPEG images are re-encoded (and limited to 2000 px on the long side); the pages stay as PDF pages. If a PDF's images are already highly compressed, or aren't JPEGs, the result may be "Original kept".
Strip the author, title, and producer fields from a PDF you're about to share externally.
Try Balanced or Strong for a PDF of scanned or photographed pages, then check the pages still read well.
Try Structure only first, since it carries zero quality risk.
Problem
Solution
The lossless pass doesn't touch images, so a scanned or photo-heavy PDF shrinks little or not at all. Choose an image option for those.
Problem
Solution
Only plain JPEG photos are. Images with transparency or masks, CMYK or colour-profile images, rotated JPEGs, tiny images and other image types are left alone and listed in the result.
Problem
Solution
Strong lowers quality to 60% and limits images to 1400 px. Open the downloaded file and check that text in scans is still readable.
Problem
Solution
Along with the size optimization, the tool blanks the PDF's title, author, subject, keywords, producer, and creator fields - re-add them afterward in another tool if you need to keep that information.
Light, Balanced and Strong each recompress the photos a little more. Stop at the first one that is small enough.
Making the images smaller before they go into the PDF gives the best control over quality.
If you just want to remove author/title/producer information from a PDF before sharing it, running it through this tool accomplishes that as a side effect.
It means nothing the tool can safely do would make the file smaller, so you keep your original.
The tool never rasterizes pages or re-renders content. It can't shrink a PDF whose size comes from embedded fonts, other image formats (such as PNG-style images), or content it doesn't recognise.
Images with a mask or transparency, CMYK or colour-profile images, JPEGs with a rotation tag, images under about 20 KB and non-JPEG images are left untouched.
Light, Balanced and Strong re-encode JPEG photos at 82%, 72% and 60% quality, and Balanced and Strong also limit their size. There is no custom quality and no target file size such as 100KB.
A password-protected PDF is refused with a message; remove the password first. Damaged files get a clear error instead of a vague failure.
Title, author, subject, keywords, producer, and creator fields are blanked as part of the optimization - this can't be selectively disabled.
The two kinds of optimization act on different parts of a PDF.
| Structure only (default) | Light, Balanced or Strong | |
|---|---|---|
| What it changes | Metadata fields and internal object structure only | Also re-encodes JPEG photos at a lower quality and, in Balanced and Strong, a smaller size |
| Quality impact | None - completely lossless | Some visual quality loss in the photos; text and vector graphics unchanged |
| Best for | Metadata-heavy or complex-structure PDFs; a safe first pass | Scanned or photo-heavy PDFs that need a large size reduction |
Which should you use?
Try Structure only first; if the file is still too large and its size comes from photos, step up through Light, Balanced and Strong until it is small enough.
The most common question is whether compression reduces quality. By default it doesn't - it only strips metadata and optimizes the internal structure. The image options do reduce the quality of JPEG photos, by the amount shown for each level. The FAQ below explains what to expect and why some files can't be made smaller.
Not by default. Structure only removes embedded metadata (title, author, subject, keywords, producer, creator) and applies object-stream compression, so text, fonts and images are unchanged. The optional Light, Balanced and Strong levels also recompress JPEG photos inside the PDF, which lowers those photos' quality by the amount shown (82%, 72% or 60%). Text and vector graphics are never changed.
It depends on the PDF. Files with a lot of metadata or many separate internal objects can shrink noticeably from Structure only. Scanned and photo-heavy PDFs shrink little or not at all with Structure only, but can shrink a lot with an image option if their pages are JPEG images. The result shows the exact before and after sizes.
Only plain JPEG photos. They are re-encoded at the chosen quality (Balanced and Strong also limit them to 2000 or 1400 px on the long side). Pages are not turned into pictures, so text stays selectable. Images with transparency or masks, CMYK or colour-profile images, JPEGs with a rotation tag, very small images and non-JPEG images are left as they are, and the result tells you how many.
If a PDF is already well-optimized, or its size mostly comes from embedded fonts, other image types or highly compressed photos, there's little the tool can safely remove. It shows "Original kept — no size reduction." in that case and doesn't offer a download, rather than giving you a file that isn't smaller.
No. There are four fixed choices (Structure only, Light, Balanced, Strong) and no target file size. Choose the lightest one that gets the file small enough.
Yes. A PDF with recompressed images is only offered if it opens again and has the same number of pages as your original, and only if it is smaller than the alternatives. Otherwise you keep the structure-only result or your original file.
A password-protected (encrypted) PDF can't be read by this tool, so it tells you to remove the password first. A file that isn't a valid PDF, or is damaged, gets a clear message instead of a vague failure. Files ending in .pdf are accepted even if your browser doesn't label their type.
No. This tool processes your PDF entirely in your browser using the client-side pdf-lib library and your browser's image tools - your file is never uploaded to any server. You can verify this yourself by opening your browser's DevTools Network tab before compressing.
Leave your email so we can prioritize similar tools and updates.