🗜️ PDF Compressor
Shrink PDF file size by compressing embedded images only: text and vector graphics stay lossless and selectable, processed locally in your browser.
Or drag a PDF file here (one file, up to 100 pages / 100 MB)
🗜️ Compression settings
Compression presetLongest image side 1600 px · JPEG quality 75%
Balances text clarity with smaller file size — ideal for mixed text-and-image reports, quotes, and PDFs exported from Word.
Only embedded images are compressed. Text and vector graphics are untouched and stay selectable after compression.
📊 Compression result
No file imported yet.
Choose a PDF file to start compressing.
👁️ PDF page preview
Pages will appear here after you choose a PDF file
How to use PDF Compressor
Follow the steps to get it done
Pick one PDF file (up to 100 pages, 100 MB)
Open the page and start — no registration or sign-in needed.
Choose the standard smart or strong compression preset and click Start compression
Settings can be changed at any time; all computation happens in your browser.
Check the size before and after, then download the compressed PDF
Copy or download the result when you are done. Nothing is kept on this site.
Page Actions
Click to pick one PDF file on your computer (up to 100 pages, a single file no larger than 100 MB).
Drag a PDF file straight onto the upload area and drop it to import it.
After upload, the upload area becomes a file info card (file name, size, and page count). A line of small text stays below the card reading “Click or drag a new file here to replace it” (with a translucent overlay appearing on hover). Clicking anywhere on the card or dragging in a new file replaces the current file and clears the previous compression result.
After a PDF is imported, thumbnails of all pages render below automatically (at low resolution), each with a page number badge at its top right, and the top shows “N pages”. Importing a different file clears the old thumbnails and renders them again.
Drag the slider, click − / +, or type a value from 50–150 directly in the number box (out-of-range values are clamped automatically and a note appears in the status area) to scale the thumbnails between 50% and 150%. It only affects how they are displayed, not the compression. The zoom value is saved with the settings cache and kept after a refresh.
Click any thumbnail and a detailed large view opens below as “Page N preview”, so you can check the page content before compressing.
A summary line stays below the thumbnail area: how many pages there are, the thumbnail zoom, and which page you are looking at. Pages that failed to render are noted as skipped.
Click the ✕ at the right of the detailed preview's title bar to fold the large view away and return to the thumbnail list.
Two pill buttons switch the preset: standard smart (images capped at 1600 pixels on the longest side, JPEG quality 75%, the default) balances sharpness and size, while strong (1024 pixels, 55%) compresses harder and suits cases with strict size limits.
Below the preset, the longest side and JPEG quality for the current preset are shown live (read-only display; v1 does not offer custom values). They change together when you switch presets.
In standby it reads “🗜️ Start compression” and processes the PDF's embedded images page by page at the current preset. It is disabled when no file is imported. When compression finishes, the same button turns into “⬇️ Download compressed PDF”. Click it to download the compressed result (the file is named “original-name-compressed.pdf”, and you can click it repeatedly to download again. The download link is destroyed right after use, so it takes no memory). Importing a different file or switching presets expires the result, and the button returns to “Start compression”.
While compressing, a progress bar and the text “Processing page N/M, compressed X images, skipped Y” show the progress live.
Click this button during compression to stop the pages that have not started yet. Pages already processed are not rolled back, and you can start again after cancelling.
When finished, it shows the size before compression, the size after, and the percentage saved, plus the count “Processed X images, skipped Y”. When images are already low quality and there is nothing to compress, it says so honestly and does not inflate the numbers. The download entry is the main button in the settings panel (it turns into “⬇️ Download compressed PDF” when finished).
A status line stays at the bottom of the compression result panel. Import, compression progress, cancellation, and failure are all written there and never disappear. Failures are marked in red with the reason.
How to Let AI Use It
Follow the steps to operate the browser page — no API key needed
There is no MCP tool for this one. Let an AI assistant follow these steps to operate the browser page (button IDs and labels match the workbench above):
- Open the tool page https://tools.waoww.com/tools/pdfcompress/
- Click the “📁 Choose a PDF file” button (#pc-selectBtn) to upload a PDF, or drag one straight onto the upload area (#pc-dropzone)
- In the preset area (#pc-modePills), pick “Smart standard” (longest side 1600 pixels, JPEG quality 75%) or “Strong compression” (1024 pixels, 55%)
- Click the “🗜️ Start compression” button (#pc-compress) to start. The progress bar shows the current page and the number of images compressed so far
- When finished, the same button turns into “⬇️ Download compressed PDF”. Click it to download the compressed result. The sizes before and after, and the percentage saved, appear in the result area (#pc-result); check the conclusion on the status line (#pc-status)
Example prompt (send it to your AI assistant as-is)
「Open tools.waoww.com/tools/pdfcompress/, compress this PDF with the standard smart preset, and tell me how much size it saved」
How It Works & Highlights
How it works, and where it fits
This tool uses pdf-lib to parse the PDF locally in the browser, walking each page's resource dictionary to find the image objects (XObjects), and only downsamples and re-encodes the embedded images as JPEG: JPEG images are decoded from their original stream and re-compressed directly, while images in lossless formats are decoded to pixels and re-encoded. Text, vector graphics, tables, and hyperlinks are kept exactly as they are, so they can still be selected and copied after compression. When the same image is shared across several pages, it is compressed only once. After replacement, the original image objects are deleted before the file is saved and packed (otherwise the old images would still take up space), and object streams are used to slim the structure further. Images that cannot be processed (JPEG2000, CCITT fax encoding, CMYK, indexed colour, JPEGs with a transparency mask, and oversized images over 50 megapixels) are skipped and counted in “skipped Y” rather than failing silently. Images nested inside form objects (Form XObjects) are not handled in v1. All processing happens locally, with no upload.
Highlights
- ✓ Only the images are compressed, never the text: vector text is untouched and still selectable after compression
- ✓ Two presets, standard and strong, covering both scanned documents and mixed text-and-image PDFs
- ✓ Runs entirely locally, so sensitive files like contracts and quotations are never uploaded
Tips
- 💡 A mixed text-and-image PDF exported from Word works well on the standard preset; a long report with many images usually saves 40%–70%
- 💡 Switch to the strong preset when a WeChat or email attachment size limit applies, after checking that the image sharpness is still acceptable
- 💡 After compressing, open the PDF and spot-check one or two pages of images to confirm the sharpness and that text is still selectable before sending it to a customer
FAQ
Privacy, format limits, and usage tips