Compresseur PDF

A PDF file is essentially a container that holds text, fonts, images, vector graphics, and metadata. When a word processor like Microsoft Word or Google Docs exports a document to PDF, it includes not only the visible content but also a significant amount of structural overhead — unused font glyphs, duplicate resource dictionaries, verbose cross-reference tables, and metadata such as the author's name, creation date, and editing history. This overhead can account for 20-40% of the total file size on documents produced by office applications.

Scanned PDFs are a different story. When you scan a document, each page is captured as a full-page image (typically JPEG or CCITT Group 4 compressed) and embedded into a PDF wrapper. In this case, the image data dominates the file size — often 90% or more — and the PDF structure overhead is negligible. This is why structure-only compression has minimal effect on scanned PDFs.

Lossless compression reduces file size without altering the visual appearance of the document. It works by removing redundant data within the PDF structure: unused objects are deleted, cross-reference tables are optimized, content streams are re-compressed with more efficient algorithms, and metadata is optionally stripped. The output is visually identical to the original — every pixel, every character, every vector path is preserved exactly. Our tool currently uses lossless compression exclusively.

Lossy compression takes a more aggressive approach. It re-encodes embedded images at lower resolutions, downsamples high-DPI photos to screen resolution, converts color images to grayscale, and may even re-rasterize vector text. Lossy compression can achieve 70-90% size reduction on image-heavy PDFs, but the visual quality is degraded. Adobe Acrobat's 'Optimize PDF' feature and Ghostscript's `-dPDFSETTINGS` options offer lossy modes.

Our compressor uses the pdf-lib JavaScript library to parse the PDF's internal structure directly in your browser. It reads the cross-reference table to locate every object in the file, identifies unused or orphaned resources, and then rewrites the document with a streamlined structure. The entire process happens in your browser's memory — no data is transmitted over the network.

Cette approche a deux avantages clés sur les outils basés sur serveur comme ILovePDF et SmallPDF. Premièrement, vos documents ne quittent jamais votre appareil — essentiel pour les documents confidentiels. Deuxièmement, il n'y a pas de limites de taille de fichier, de limites quotidiennes ni de files d'attente.