PDF Compressor

Photos usually make up most of a PDF's size: keep the text and re-save only the photos to shrink it a lot

The PDF Compressor shrinks a PDF by re-saving the photos inside it at a smaller size and tidying up pieces the file no longer uses. The default method, Shrink images only, leaves text, tables, links and bookmarks untouched, so you can still search and copy sentences after shrinking. The tool's sample report (3 pages, 2 photos) goes from 2.0 MB to 384 KB at Balanced quality, 80.7% smaller.

Once you choose a file, bars show how the original's size breaks down (images, fonts, page content, other), and a result bar shrinks next to it. Pick an upload limit and a limit line is drawn too, so you can see at once whether the result fits. The original and the result sit on top of each other on the same page, and you drag the divider to compare them. The PDF is never sent to a server: it is read and saved only inside this browser.

When you need to shrink a PDF

A text-only PDF stays at a few hundred KB, but a phone photo or a scanned page can add several MB to a single page. Size becomes a problem mostly when you upload or send the file somewhere.

• Attaching it to an email: Personal Gmail accounts allow attachments up to 25 MB, and when the total goes over that, Gmail replaces the attachment with a Google Drive link. Outlook.com also limits attachments to 25 MB.
• Uploading to an application, form or assignment portal: each upload field sets its own limit for a single file. Pick the limit from the instructions in the upload limit menu and you can see right away whether the result fits.
• Sending a PDF built from photos: documents made with a scanning app or a photo-to-PDF converter often carry the photos at full original size. Shrinking them to a size that still reads well on screen cuts the file dramatically.
• Passing it around on a phone or in a messaging app: a small file downloads and previews faster and uses less data and storage.

For documents that carry a digital signature, such as contracts and certificates, or manuscripts headed to a print shop, check the recipient's requirements before shrinking. A shrunk file is a re-saved copy, so signature verification and print quality are not guaranteed.

How to use it

1. Drop or choose a PDF. Drag it onto the box or press [Choose PDF]. With no file at hand, [Shrink the sample] runs a made-up 3-page report.
2. Read the result size. As soon as a file opens, it is shrunk with the Standard compression level (Shrink images only · Balanced), and you see the result size, the percentage saved and the original and result breakdown bars.
3. Pick an upload limit. Choose from 1, 2, 5, 10 or 20 MB and a limit line appears on the bars. If the result is over, the tool says by how much and suggests the next method to try.
4. Pick a compression level. Press one of the six levels, Lossless, Light, Standard, Strong, Very strong or Maximum, and the file is recalculated at that level. Each computed size stays under its level so you can compare them. The hatched Very strong and Maximum levels turn pages into images, so text search is lost. To choose the method and quality separately, open [Choose method and quality yourself].
5. Check in the overlay and save. Choose a page from the page weight bars or the page buttons, then drag the central divider to compare the original with the result. The overlay fits the whole page on screen, and [Zoom] doubles the size so you can read small print. When you are happy, press [Save].

How the three methods differ

Cleaning up the structure doesn't change the content, so it saves little. For a document with photos, Shrink images only saves far more, and the text stays real text. Pages to images packs fonts and shapes into a single picture and is often the smallest, but the text in the result is text inside a picture: you can't search or copy it, and no OCR (text recognition) is applied.

Reading the size breakdown bars

The original bar splits the file's bytes into four parts. The four parts add up to the file size, and the result bar is split the same way. Seeing which part is large tells you in advance which method is likely to help.

• Photos and images: the photo, scan and picture files inside the document. This is the part Shrink images only and Pages to images reduce. The count includes the transparency data (mask) of a see-through image as one.
• Fonts: the font files embedded in the document. Shrink images only leaves fonts as they are, so in a text-heavy document this part stays and the file doesn't shrink much.
• Page content: the commands that place text and draw lines and shapes. It grows with drawings and complex charts.
• Other: everything else that forms the document's structure, such as the page list, bookmarks, links and document info. This is mainly what Clean up structure only reduces.

The page weight bars below show, for each page, the size of its content plus the photos used there for the first time. Fonts shared by several pages aren't assigned to any single page. Each page gets one row: the dark bar is the current size and the light bar is what was saved. Press [Heaviest first] to bring the heaviest pages to the top. An unusually long bar means the photos on that page take up most of the size, and if the page isn't needed, removing it is the surest way to shrink the file.

The real values behind the three qualities, and example results

Quality applies to both Shrink images only and Pages to images. Shrink images only scales a photo's long side down to the number of pixels an A4 long side (297 mm, about 11.69 inches) would have at that dpi. If the photo is already smaller, its size is kept and only the JPEG quality is applied again. If the re-saved photo isn't at least 10% smaller than the original, the original photo is kept.

JPEG quality is a value between 0 and 1 handed to the browser's JPEG encoder. 0.72 doesn't mean 72% of the original size or 72% of the original sharpness, and the same value can give slightly different sizes in different browsers. Below are the results of shrinking the sample report (1,994,250 bytes: images 1.8 MB, fonts 148 KB, page content 5 KB, other 12 KB) in Chrome.

Even after Shrink images only, the sample's 148 KB of fonts stay. That is why Pages to images, which packs the fonts into the picture, came out smaller, at the price of text you can no longer search. For scanned documents where photos weigh even more, Shrink images only alone often gets close to the same size.

Upload limits and how MB is counted

This tool counts 1 MB as 1,000,000 bytes and 1 KB as 1,000 bytes. Some upload fields count 1 MB as 1,048,576 bytes (1 MiB), and then the limit is about 4.9% more generous than this tool's. A file that comes out within a limit here is therefore within it under either way of counting.

The percentage saved is calculated as (original bytes − result bytes) ÷ original bytes × 100 and shown to one decimal place. It uses the actual bytes before the on-screen MB values are rounded, so subtracting the numbers beside the bars may differ in the last digit. Whether the limit is exceeded is also decided on actual bytes: a result shown as 10.0 MB is reported as over a 10 MB limit if it is more than 10,000,000 bytes.

If the result isn't smaller than the original, the save button for that method is turned off. Re-saving a document that is already well compressed can make it larger, and the tool won't have you download such a file as if it had been shrunk.

What to check in the overlay

The overlay draws the original and the result of the same page at the same size, one on top of the other. Left of the divider is the original and right of it is the result. Drag the divider or click the view to move it, or focus the divider and press the left and right arrow keys.

• Small print and numbers in tables: on scanned, photo-like pages, press [Zoom] to double the size and see whether the digits are still readable.
• Stamps, signature images and thin lines: faint stamps and pencil lines can smear when re-saved as JPEG.
• Photo colors and edges: if banding in steps appears in even areas such as sky, choose the next sharper quality.
• The heavy page, not just page one: sort the page weight list by [Heaviest first] and click a row with a long bar to look first at the page that actually shrank the most.

Documents that don't shrink well

• Text-only documents: most of the size is fonts and page content, so Shrink images only barely changes anything. Clean up structure only may save a little, or it may be better to leave the file alone.
• Scans already compressed at low quality: if the photos are already small and the JPEG quality is low, re-saving doesn't save more than 10%, so the original photos are kept.
• Black-and-white scan formats: this tool doesn't re-save images stored with compression meant for black-and-white documents, such as JBIG2 and CCITT fax, or in JPEG 2000 (JPX). Those formats are usually already stored small.
• CMYK photos for print and spot-color images: only RGB and grayscale photos meant for screens are re-saved. CMYK and indexed-color images, as well as grayscale photos tagged with a gray ICC color profile, are left as they are so the colors don't change.
• Drawings and complex charts: line and shape commands (page content) take up the size, so if you must shrink one, try Pages to images and zoom in to check the lines stay readable.

Common mistakes

• Shrinking an already shrunk file: applying Shrink images only again and again re-saves the photos as JPEG each time, so quality keeps getting worse. To shrink further, go back to the original and choose the next smaller quality.
• Deleting the original and keeping only the compressed copy: a file turned into page images can't get its search, copy and form fields back. Keep the original somewhere.
• Submitting a shrunk signed contract: for a PDF with a digital signature, the verification breaks the moment it is re-saved. If the tool finds a signature, it tells you when you open the file.
• Searching for a sentence after Pages to images: the text in the result is text inside pictures. Use Shrink images only for documents you need to search.
• Going by the limit number alone: check the instructions to see whether the upload field sets its limit per file or as a total across files.

Limits and cautions

• One PDF at a time, up to 100 MB. Pages to images works up to 300 pages. A big file can stall a phone for lack of memory, so process it on a computer.
• PDFs with an open password or an editing-restriction password can't be shrunk. If you have the right to the document, try again with an unlocked copy.
• The tool doesn't guarantee any particular size. If you're over the upload limit, try the next method the screen suggests, and if it's still too big, remove pages you don't need or re-export from the original program with lower photo resolution.
• Transparency data (masks), images smaller than 8 KB and large images over 50,000,000 pixels are left as they are.
• The first time you use the tool, it downloads the processing engine (pdf-lib 1.17.1, pdf.js 3.11.174, about 1.9 MB) once. The PDF itself is not sent to a server, but loading the site's pages and ads involves separate network traffic.
• The sizes in the example table were measured in Chrome. Other browsers make JPEGs differently, so result sizes may differ.

Checked on The standards, help pages and example results were checked on 2026-10-20.

Sources

• 01ISO 32000-1:2008 Document management — Portable document format (Adobe public copy)7.5.7 object streams from PDF 1.5, Table 6 FlateDecode = zlib/deflate compression, 7.4.8 DCTDecode = JPEG baseline image data
• 02Gmail Help — Attach files in GmailAttachment limit of 25 MB for personal Gmail accounts; larger files are added as a Google Drive link
• 03Microsoft Support — Sending limits in Outlook.comOutlook.com attachment size limit of 25 MB
• 04MDN — HTMLCanvasElement.toBlob()image/jpeg quality is a value from 0 to 1; the browser default applies if it is omitted
• 05MDN — createImageBitmap()Reading a photo at a smaller size with resizeWidth, resizeHeight and resizeQuality
• 06MDN — <canvas>: Maximum canvas sizeCanvas area limit of 4,096 × 4,096 pixels on iOS devices (applies to Pages to images and the overlay)
• 07pdf-lib — library for creating and editing PDFsMIT license. This tool reads and re-saves PDFs with version 1.17.1
• 08pdf.js — Mozilla's PDF rendering engineApache-2.0 license. This tool draws the overlay and Pages to images with version 3.11.174

Frequently asked questions

Tools to use with this one