What does PDF Compress do?
It builds a new PDF using either raster page images or a structure-only page copy, then downloads that separate compressed file.
Pick a compression preset, review before-and-after sizes, and inspect the new PDF before you share it.
Preparing the editor...
The guide and instructions on this page are available while the tool loads.
PDF Compress creates a new PDF that may be smaller than the source, but savings depend on how the original was built. Image-heavy scans often respond differently from text-focused digital documents, and some files change only a little or grow larger.
Maximum, Balanced, and High quality render each processed page with pdf.js, optionally convert the canvas to grayscale, encode it as JPEG, and rebuild the PDF from those images. Repack only copies the selected pages into a new document and saves with object streams, without intentionally rasterizing page content or recompressing embedded images in place.
Raster presets turn text and vector artwork into pixels, so selectable or searchable text and interactive page features such as forms and links are not preserved on those pages. Page dimensions follow the rendered viewport size. Repack only keeps pages as PDF content and can retain selectable text, but document-level bookmarks, attachments, metadata, and every interactive feature should still be checked after download.
All compression work stays in the browser. The PDF is not posted to a DevToolKit compression API. Ordinary website assets and the pdf.js worker may still load; the compress workflow itself does not emit identified tool analytics for filenames, PDF bytes, or document text.
The tool accepts one PDF up to 50 MB. There is no password field and no password-removal feature. Creating a new file typically invalidates certificate-based signatures. If a raster result stays at about 98% of the original size or larger, the tool may try a structure repack and keep the smaller of the two attempts. There is no dedicated cancel control once compression starts.
Typical tasks this tool is built for.
Compression runs locally in your browser with pdf.js, canvas JPEG encoding, and pdf-lib. The selected PDF is not uploaded to DevToolKit for compression, and the PDF Compress screen does not send identified tool analytics events for the file. Ordinary site assets, the local pdf.js worker, and general site activity may still load.
Practical tips before you download or share the output.
PDF Compress accepts one PDF up to 50 MB with no separate hard page-count cap. Default preset is Balanced. Maximum, Balanced, and High quality rasterize processed pages to JPEG at fixed render scales and quality values; grayscale is available for raster modes and is enabled by Maximum. Repack only copies pages with object streams and does not downsample embedded images. Output is always a separate {base}-compressed.pdf. Size can shrink, stay similar, or increase. A positive advanced page limit processes only the first N pages. Raster results lose selectable text and interactive page behavior. Source metadata is not intentionally copied. Encrypted PDFs may fail; this tool does not remove passwords. Rewriting typically invalidates certificate signatures. Automatic and manual download are available. Reset clears the workspace and does not abort in-flight work. Large raster runs can use substantial browser memory.
Longer reads that pair well with this tool.
PDF guide
Match your PDF task to the right DevToolKit tool — merge, split, compress, convert, secure, or edit — and check processing labels before you upload.
PDF guide
Step-by-step In-browser PDF compression: raster downsampling, structural repacking, and how to treat scanned versus text-heavy files — with no quality guarantee.
PDF guide
Practical student workflows — compress scans, merge assignments, split chapters, convert phone photos to PDF, and protect personal records — with honest limits on OCR output.
Common next steps after using this tool.
Combine whole PDF files into one document that follows the order shown in your file list.
Encrypt one PDF with a required open password, optional owner password, and viewer permission presets or custom print, copy, edit, and annotate controls.
Build an editable DOCX from the PDF text layer, with optional Page N headings and best-effort Word tables.
Turn PDF text into an editable XLSX workbook with page sheets, detected-table sheets, or one combined data sheet.
Recognize printed text on scanned or image-based PDF pages locally, then export a TXT or Markdown working copy for proofreading.
Select pages and ranges from one PDF and copy them into one new extracted PDF downloaded directly.
Situations where this workflow saves time.
It builds a new PDF using either raster page images or a structure-only page copy, then downloads that separate compressed file.
One PDF at a time.
One PDF up to 50 MB. There is no separate hard page-count limit, but large raster jobs can need substantial browser memory.
Local. The workflow uses pdf.js, canvas JPEG encoding, and pdf-lib in the browser.
No. The selected PDF is not posted to a DevToolKit compression API. Ordinary site assets and the local pdf.js worker may still load.
Maximum, Balanced, High quality, and Repack only. Balanced is the default.
Maximum uses scale 1 and JPEG quality 0.52 with grayscale on. Balanced uses scale 1.25 and quality 0.72 in color. High quality uses scale 1.55 and quality 0.88 in color.
It copies selected pages into a new PDF and saves with object streams. It does not intentionally rasterize pages or recompress embedded images in place.
No. You choose a preset and optional grayscale or page limit; the tool does not chase an exact kilobyte target.
No. Results vary by source content. The UI reports original size, new size, and whether savings were meaningful.
Rebuilding pages as JPEG images or rewriting structure can add overhead when the source is already efficient. The tool still returns that result for you to compare.
Image-heavy or scanned pages often respond more to raster presets. Text-focused digital PDFs may gain more from Repack only, or they may show little change.
No. Raster presets turn each processed page into a JPEG image, so text is no longer selectable or searchable as PDF text.
Processed pages keep source order. With page limit 0, all pages are included; a positive limit keeps only the first N pages.
Repack only copies existing pages and can keep their sizes. Raster mode creates new pages sized to the rendered viewport, so dimensions can change.
Raster mode removes interactive behavior by using page images. For Repack only, verify important forms, links, and annotations in the downloaded file.
Document-level bookmarks and attachments are not intentionally copied into the new PDF.
There is no password input. Encrypted files may fail during inspection or compression. Use PDF Unlock on an authorized copy first.
Creating a new compressed PDF typically invalidates certificate-based signatures. Keep the signed original.
The download is named {original-base}-compressed.pdf. The original file stays unchanged.
There is no dedicated cancel control after Compress PDF starts. Reset clears the workspace afterward but may not stop work already running.
The PDF Compress screen does not call identified tool analytics for the file. Ordinary website requests can still occur, and PDF bytes are not uploaded for compression.
Browse related tools or open the full workspace.