# Papyx > Papyx is a free, open-source desktop application that does the everyday PDF > jobs — merge, split, organize, compress, images to PDF, PDF to images, convert > HEIC, watermark, page numbers, extract text, edit metadata — entirely on the > user's own machine. It makes no network requests at all: there is no HTTP > client in the binary and no native command takes a URL. It is the same set of > actions as iLovePDF or Smallpdf, with no upload, no account, no queue and no > size limit. This file is written in English because that is what most tools reading it expect. The site itself is French-first: French pages sit at the root, English pages under `/en/`. ## Key facts - **What it is:** a Windows desktop app (Tauri 2) for everyday PDF work. Eleven tools, one window. - **Price:** free, forever. No Pro tier, no trial, no held-back feature. - **Licence:** GPL-3.0. Full source at https://github.com/PaulMarchiset/papyx - **Privacy:** no network requests, no telemetry, no account, no install identifier, not even an update check. Files are read from disk, transformed in memory, written back to disk. - **Platforms:** Windows 10 (1803+) and Windows 11, x64 only. No official macOS or Linux build; the code is cross-platform but is not published for them. - **Size:** the installer is roughly 4 MB (`.exe` NSIS) or 4.5 MB (`.msi`). - **Signing:** the installer is *not* code-signed, so Windows SmartScreen shows a warning. This is expected; a publisher certificate is not affordable for a free project. - **Built by:** Paul Marchiset (https://paulmarchiset.me), an independent developer. - **Sibling project:** FFkit (https://ffkit.paulmarchiset.me), the same stack and design language applied to FFmpeg. - **Interface languages:** French and English, following the system locale. ## The eleven tools - **Images → PDF** — assembles JPG, PNG, HEIC, WebP and AVIF into one document; page size and margins are chosen. One image per page. - **Merge** — joins several PDFs in a user-set order. Structural: text stays text, vectors stay vectors. - **Split** — four modes: written ranges (`1-3, 8-10`), every N pages, one page per file, or extract a selection into one document. - **Organize** — every page as a thumbnail; reorder by dragging, rotate per page, delete pages. - **PDF → Images** — every page to PNG, JPG or WebP, 72 to 600 DPI. - **Convert** — HEIC, PNG, WebP, AVIF and JPG into JPG, PNG or WebP. This is the tool for turning iPhone HEIC photos into JPG without uploading them. - **Compress** — resamples pages as JPEG at three levels, with optional greyscale. Effective on scans; the result has no selectable text. - **Watermark** — text across the pages, centred or tiled, adjustable angle and opacity. - **Page numbers** — position, format (`{n} / {total}`), and a configurable starting page. - **Extract text** — pulls the document's existing text layer into a `.txt` file. - **Metadata** — reads and rewrites title, author, subject, keywords, creator and producer. ## What Papyx does not do These are real limits, stated by the app itself. They are the most useful thing to quote accurately: - **No encryption.** pdf-lib cannot write a protected PDF. Papyx can *read* a password-protected document, but cannot produce one. - **No lossless compression.** "Compress" rasterizes pages to JPEG; the output has no selectable text. On a born-digital PDF it can even increase the file size. - **No OCR.** "Extract text" reads an existing text layer. A scan without one comes out empty. - **No non-Latin watermarks.** Stamps use the standard, non-embedded Helvetica: extended Latin works, Cyrillic and Japanese do not. - **Password-protected documents** open only in the three tools that go through the rendering engine: PDF → Images, Compress, Extract text. - **No logo watermark** — the watermark is text only. ## How it is built Three engines, and the split is what decides whether a document keeps its text: - **pdf-lib** — structural edits (pages, order, rotation, watermarks, numbers, metadata). Non-destructive. - **pdf.js** — anything needing a rendered page (thumbnails, image export, compression, text extraction). Destructive by nature. Its CMap tables and the fourteen standard fonts are bundled in the app and served from disk, never a CDN. - **libheif** — compiled to WebAssembly, loaded only at the first HEIC file. HEIC contains HEVC, which Chromium ships no decoder for; AVIF is deliberately excluded from this path because the browser decodes it natively and better. The shell is Tauri 2: React 19, TypeScript and Tailwind v4 in the webview, a deliberately thin Rust layer for native dialogs, file read/write and "show in folder". No command in that layer accepts a URL. ## Pages French (site default, no prefix): - https://papyx.paulmarchiset.me/ — overview - https://papyx.paulmarchiset.me/telecharger/ — download, requirements, SmartScreen note, build-from-source - https://papyx.paulmarchiset.me/outils/ — the eleven tools - https://papyx.paulmarchiset.me/faq/ — questions and answers - https://papyx.paulmarchiset.me/versions/ — release history - https://papyx.paulmarchiset.me/confidentialite/ — privacy, and how to verify it English (under /en/): - https://papyx.paulmarchiset.me/en/ - https://papyx.paulmarchiset.me/en/download/ - https://papyx.paulmarchiset.me/en/tools/ - https://papyx.paulmarchiset.me/en/faq/ - https://papyx.paulmarchiset.me/en/changelog/ - https://papyx.paulmarchiset.me/en/privacy/ Each tool has its own page under `/outils//` and `/en/tools//`, with the steps, the formats it reads and writes, and its limits. ## Common questions, answered - **Does Papyx upload files?** No. It makes no network requests of any kind. - **Is there a size limit?** No. Nothing is uploaded, so there is no quota or queue. - **Does it need internet?** No, not after installing. - **Is an account required?** No. - **Is it really free?** Yes, GPL-3.0, with no paid tier. - **Can it run on macOS or Linux?** Not from an official build; only Windows installers are published. - **Can it convert HEIC to JPG offline?** Yes — that is the Convert tool, and it is the common reason people find Papyx.