What is PDF Compress?
PDF compression is more nuanced than image compression. A PDF file isn’t just a collection of images — it contains text streams, vector graphics, fonts (sometimes fully embedded, sometimes subsetted), form fields, metadata, and embedded thumbnails. Each component requires a different compression strategy. This tool uses a WebAssembly port of a Ghostscript-equivalent compression pipeline to process your PDF locally. The compression targets the components that account for most of the file size: embedded images are downsampled, unused font glyphs are stripped, redundant embedded thumbnails are removed, and metadata is reduced. Text and vector graphics remain crisp because they are lossless by nature. **Why your file might get bigger** — If your PDF already uses JPEG2000-compressed images (common in PDFs exported from InDesign or high-end scanners), re-encoding them as standard JPEG can sometimes increase file size. This is not a bug — it’s the trade-off between universal compatibility and compression efficiency. **Compression levels** — Choose from screen (smallest, ~72 DPI images), ebook (medium, ~150 DPI), printer (high quality, ~300 DPI), and prepress (maximum quality). For email attachments and web delivery, ebook is usually the right choice.
A note about file privacy
PDF Compress is built to handle your file entirely in the browser. You can confirm the data path in DevTools: during processing, your file should not show up as a network upload request. For the broader risks of fake or untrusted converters, see theFBI Internet Crime Complaint Center warning.
Treat PDF Compress like a small desktop utility, not an upload service. Your browser may fetch the code needed to do the work, but the selected file stays in local memory while it is processed. That is why the Network panel is worth checking whenever the file is confidential.
- Before processing: strip metadata, comments, and form fields you would not want shared, since these can carry author names, paths, and internal notes into the output.
- While processing: watch the Network tab. A library download is expected; a request carrying your file bytes is an upload.
- After downloading: scan unfamiliar results before opening them. A file that looks converted can still be malicious.
Supporting guidance: Malwarebytes on malicious converters andKaspersky's safe conversion guidance.
Deep Dive: PDF Compress
Related Articles
Learn more about this tool and related topics in our blog.
Why Developers Prefer Offline File Tools in 2026
Privacy isn't a perk, it's a requirement. See why top developers are ditching cloud converters for local-first browser utilities.
How Browser-Based File Tools Work (WebAssembly Explained)
Peek under the hood of Filemint. A practical look at WebAssembly, Web Workers, and the browser APIs behind our private file tools.
How to Process Files Privately Without Uploading Them
Your files stay on your device. This guide explains how Filemint processes them in the browser instead of sending them to a server.
“”
Azeem Mustafa
Privacy Architect
Core Capabilities
- Adjustable image compression quality
- Local WebAssembly processing for maximum speed and security
- Preserves original layout, text searchability, and links
- Completely free with no limits on document page counts
- Works offline without sending data to servers
Why It Matters
- Privacy: Process confidential, financial, or legal documents safely.
- Speed: Instant local compression without waiting for server uploads or downloads.
- Efficiency: Dramatically shrink files for email attachments and web portals.
- Control: Choose your preferred quality level to balance size and clarity.
Quick Start Guide
Upload your PDF: Drag and drop your PDF file or click the upload button to select it.
Choose compression level: Select a target JPEG image quality slider. 50-70% gives the best balance of small file size and clear graphics.
Compress: Click the "Compress PDF" button to trigger the local optimization process.
Download your optimized PDF: Once done, click download to save the compressed PDF to your device.
Usage Examples
Image heavy report to 1.2MB
Scenario 01A 5MB marketing report with product photos shrinks to about 1.2MB at 70 percent quality, small enough for email.
product_report_5mb.pdf
product_report_compressed_1.2mb.pdf (76% smaller)
File that grew from font embedding
Scenario 02A 2MB text PDF with one logo re-encodes to 2.4MB because the source used lossless PNG and the tool writes a full embedded font program.
notes_2mb.pdf
notes_compressed_2.4mb.pdf (19% larger, fonts added)
Metadata stripped scan
Scenario 03A 9MB scanned contract drops to 2.1MB after image re-encoding and object stream packing remove old revision data and padding.
scan_contract_9mb.pdf
scan_contract_compressed_2.1mb.pdf (77% smaller)
Email attachment fix
Scenario 04A 14MB signed agreement with photos becomes 1.6MB, clearing a 10MB mailbox limit with room to spare.
agreement_14mb.pdf
agreement_optimized_1.6mb.pdf (89% smaller)
Common Scenarios
Portal submission limits
Uploading a document to a portal with a 5MB maximum.
Confidential legal filings
Compressing legal briefs without cloud leakage risk.
Scanned research paper
A 300 DPI scan that is too big to email.
Monthly invoices
A stack of invoice PDFs with logos and charts.
Image heavy annual report
A report full of photographs and charts.
Email attachment cleanup
A contract draft too big for a mailbox cap.
Long term archival
Storing old paperwork without wasting disk.
Design portfolio PDF
A portfolio of high resolution artwork.
Questions?
Technical Architecture
Re-encoding embedded images
The engine scans for Image XObjects, decodes each pixel stream, and re-encodes it. Photos become tighter JPEG data, which is where most of the savings come from. PNG sources may move through a quantization path that drops colors for a smaller result.
Object stream compression
Beyond images, the tool packs dictionaries and other indirect objects into compressed object streams using Flate compression. This removes the plain text overhead that older PDF writers leave behind and drops dead data from prior incremental saves.
Font embedding tradeoff
Fonts are preserved so layout stays correct. If the source referenced system fonts, embedding a full font program can add bytes. That is a known reason a text heavy file may not shrink or may even grow.
Browser memory ceiling
WebAssembly on 64 bit browsers can address up to about 4GB of linear memory. The decoded images plus the file buffer must fit inside that working set, so the practical file limit is well below 4GB of source size.
Local WASM pipeline
All decoding, re-encoding, and PDF rebuilding happen in a WebAssembly module running in the tab. No bytes leave the device, which is why there is no upload step and no server queue.
DPI aware downsampling
When the user lowers the DPI setting, embedded images are resampled before re-encoding. Dropping a 300 DPI scan to 150 DPI roughly cuts the pixel count by four, which is the single biggest lever for scanned documents.
Why PDFs bloat in the first place
A PDF carries more than the words you see. Every photograph is stored as its own image, often at the full resolution of the scanner or camera that made it. A single 300 DPI color page can hold tens of megabytes of image data. On top of that, fonts, bookmarks, and metadata dictionaries sit alongside the pages, and each time a file is saved with incremental updates, old versions linger inside the container.
That is why two files with the same page count can differ wildly in size. A text report might be 200KB, while a scan of the same report is 20MB. The compression job is really about taming the images and clearing out the leftover structure, not about rewriting the text. If you also work with loose images, our image compressor follows the same private approach.
Real world ranges from a 2026 browser compression study. Your results vary with image density and chosen quality.
Picking the right quality
The quality slider is the main control. Around 85 percent keeps photos looking untouched to the eye while still cutting size. Drop to 60 or 70 percent when you only need the file to fit an email or a portal limit and on screen reading is enough. Pair a lower setting with 150 DPI for scans and you will see the biggest drops. When a document must stay print sharp, keep quality high and only trim the DPI a little.
If privacy is the reason you are here, the guide on client side processing and privacy explains why keeping files on your device matters. After compressing, you can also strip leftover author and producer data with the metadata remover. For the broader mechanics of how PDFs are built and taken apart, our PDF transform mechanics guide goes deeper.
Average size drop
Across common document types
Browser WASM memory cap
On 64 bit desktop browsers
Bytes uploaded
Everything stays on device
DPI sweet spot
For scans sent by email
How FileMint compares with popular server based tools.
| Feature | ★ RecommendedFileMint | Smallpdf | iLovePDF |
|---|---|---|---|
| Files stay on your device | |||
| Free daily task limit | |||
| Upload required | |||
| Max free file size | 100 MB | 200 MB | |
| Works offline |
Build a private PDF workflow
Compression is one step in a larger routine. You might merge several scans into one file, split a huge archive into email sized chunks, compress the result here, then strip metadata before sending. Each tool in that chain runs locally, so the document never leaves your control. If you want to know which image format saves the most for future scans, the best image formats guide breaks down JPEG, WebP, and AVIF tradeoffs.
The facts behind this page draw on several public references. The PDF format itself is defined by ISO 32000-2 (PDF 2.0). The PDF Association explains object streams and redundant structures in its PDF optimization writeup. For the image codecs we rely on, see MozJPEG and OxiPNG. The browser memory ceiling is covered by the V8 team's 4GB WASM note, and the library we use to read and write PDFs is pdf-lib.
Keep Exploring
Power up your workflow with related utilities.
Related Tools
PDF Watermark
Add custom text watermarks to your PDF documents. Adjust opacity, rotation, size, and position. The processing happens in your browser memory, so your documents are never uploaded to any server.
Use free →Protect PDF
Protect a PDF with a user password before sharing it. This browser-only tool uses the library’s 128-bit RC4 PDF encryption, so the unencrypted file is not uploaded or exposed.
Use free →Text to PDF
Paste text, upload.txt/.md files, customize typography and layout, preview estimated pagination, and export a production-ready PDF in seconds. Everything runs in your browser with no upload required.
Use free →Related Articles
Learn more about this tool and related topics in our blog.
Why Developers Prefer Offline File Tools in 2026
Privacy isn't a perk, it's a requirement. See why top developers are ditching cloud converters for local-first browser utilities.
How Browser-Based File Tools Work (WebAssembly Explained)
Peek under the hood of Filemint. A practical look at WebAssembly, Web Workers, and the browser APIs behind our private file tools.
How to Process Files Privately Without Uploading Them
Your files stay on your device. This guide explains how Filemint processes them in the browser instead of sending them to a server.
Founder & Lead Developer at FileMint
Building privacy-first browser tools powered by WebAssembly. Focused on making file processing fast, secure, and accessible — without ever uploading your data to a server.
View full profile →