Features
Only what the build can actually do
56 verified capabilities across 12 groups. Every item is backed by the shipped test suite — features we could not confirm are simply not listed.
How to read this page
The Stop fix described below is in the published packages
PDF viewing
Open documents, browse pages with cached thumbnails, and read at the zoom you choose.
Fast page previews
Every page is rasterised once and cached on disk as a small JPEG — reopening a folder paints instantly.
Real zoom modes
Whole page, Fill width and Actual size — measured at exactly 595 × 842 px for an A4 page.
Page thumbnails
Jump through the document from a thumbnail strip in the editor.
Windows integration
Desktop and Start Menu shortcuts, .pdf file association and an “Open with Irfan PDF” menu entry.
PDF editing
Change existing text and add new content — while the page keeps looking like the original.
Edit existing text
Click any span and retype it, with font, size, colour, weight and alignment controls.
True font preservation
An edited word is written back with the document's own embedded font at its own size, weight, colour and spacing.
Add text, images and signatures
Draw, upload or type a signature; place text boxes and pictures anywhere on the page.
Shapes and annotation
Rectangles, ellipses, lines, arrows, highlights and free-hand drawing, all resizeable with 8 handles.
Undo, redo and autosave
Ctrl+Z / Ctrl+Y, plus a restorable draft if you close a document before saving.
Form-safe editing
Covers are fitted to the ink they hide, so table borders, rules and grids survive an edit (verified pixel-by-pixel).
OCR
Turn scans and photos into text you can search, select and fix — offline, with the English PP-OCRv5 model.
Searchable OCR
An invisible text layer is added over the original pixels: the page looks untouched, the words become findable.
Editable OCR
Recognised words are inserted as real, selectable, editable text at their original positions — tables included.
Auto-orientation and deskew
Sideways or upside-down scans are turned upright on upload, and tilted pages (0.3–5°) are straightened automatically.
Image to Text
OCR a photo or screenshot, correct the words on screen, and save a .txt file.
Fully offline engine
RapidOCR on ONNX Runtime with bundled models — no Tesseract install, no internet, no API keys.
Conversions
Move documents between PDF, Word, Excel and images with validation before anything is called “done”.
PDF to Word
Native layout converted with pdf2docx, using Microsoft Word itself as an accuracy fallback when installed.
Image to Word
JPG/PNG scanned pages become a paragraph- and table-aware .docx.
Word to PDF
Microsoft Word automation with a built-in converter fallback if Word is missing.
JPG / PNG to PDF
Assemble images into a PDF from the dashboard.
PDF to Images
Render every page of a document as an image file.
Extract Images
Pull the actual embedded picture objects out of a PDF, validated against their declared pixel sizes.
PDF to Excel
A real .xlsx workbook built from the page's columns — verified by opening it in Excel, where cells read as numbers.
Extract text
Save any document's text as a .txt file, with page selection.
Organise pages
Assemble, rearrange and finish documents without a second tool.
Merge
Combine several PDFs into one document.
Split
Break a document into separate files.
Rotate
Turn pages left or right in bulk.
Delete & extract pages
Remove what you don't need or pull selected pages into a new file.
Reorder, duplicate, insert
Type a page order (or reverse it), duplicate pages, and insert blank pages at front, end or after any page.
Crop and resize
Trim edges or set a custom box; resize pages to A3/A4/A5/Letter/Legal.
Remove blank pages
A page must have no text, no images and no drawings before it is called blank — sparse pages are kept.
Numbers, watermark, details
Configurable page numbers, repeated diagonal watermarks, and title/author/subject/keywords written into the file.
Greyscale, flatten, repair
Convert colour to greyscale, flatten form fields, and repair damaged PDFs.
Compression
Shrink documents and images — either pick a level or name the exact file size you need.
Quick compression
Low / Medium / High quality dials for PDFs.
Target file size
Type the size in MB: up to seven passes run, each result validated, and an unreachable target is reported honestly.
Image compressor
JPG, JPEG, PNG, WEBP and TIFF — quick levels or a target size, with the result stating if resolution was reduced.
Image Quality Enhancer (2x / 4K / 8K)
Local denoise, contrast, sharpening and Lanczos upscale — labelled in the app as NOT an AI upscaler.
Security tools
Protect what matters and remove what must not travel with the file.
Password protection
Real AES-256 encryption; the saved file is proven to refuse a blank password.
Remove password
Unlock a PDF when you supply the password; a wrong password is reported, never ignored.
Permanent redaction
Redacted characters are deleted from the file — they cannot be copied, searched or recovered (verified by re-reading the output).
Digitally signed files are refused
Instead of silently invalidating a signature, the app refuses to modify digitally signed PDFs and says why.
Watermark
Repeated diagonal text with colour, size, strength and page choice.
Search & compare
Find anything in your library and see exactly how two versions differ.
Full-text search
Search file names and document content together, with the matching sentence shown and an “IN TEXT” badge.
Compare documents
One HTML report with a word-level text diff and a page-by-page picture comparison — differences painted red.
Batch operations
Run the same job over up to 200 documents in one pass.
Seven batch operations
Watermark, compress, rotate, page numbers, greyscale, remove-blank and resize.
CSV report and per-file outcomes
Each document succeeds or fails on its own; the run writes a CSV with original and final sizes.
Stop and retry
Stop stops a batch between documents and a running OCR job between pages — a stopped job reports cancelled, never done — and one click retries only the documents that failed.
Build a real print document, then hand it to Windows.
Page ranges and N-up
Ranges like “1,3,5” or “odd” / “last”, and 2/4/6/8/9 pages on one sheet with scaling.
Print view at true paper size
Every sheet rendered at 200 DPI from the print document — A4 measures 209.9 × 297 mm on screen.
Send to printer or save as PDF
Hands the file to the Windows print dialog (which can also save with “Microsoft Print to PDF”).
Local document AI
Free analysis that runs on your computer and quotes the document instead of guessing.
PDF AI Assistant (local text analysis)
Overview, summary, key points, extracted dates/amounts/references, and page-scoped questions answered with quoted sentences.
Nothing is invented
Every sentence shown is re-read from the file and confirmed server-side before you see it — no network, no API key, no model.
PDF AI Analyst
Classifies documents, runs OCR on scans automatically, produces A-X cited analysis and exports to Word, Excel or PowerPoint.
Platform
Local-first by design, with the operational pieces teams need.
100% offline
The server binds 127.0.0.1 only; no telemetry, no analytics, no external HTTP calls exist in the code.
Optional multi-user
Sign-in is off by default; switch it on for ADMIN/USER accounts with per-tool permissions and PBKDF2-hashed passwords.
Health check / self-test
20+ checks that each execute a real operation — OCR, conversions, encryption — with an exportable diagnostic report.
Light and dark themes
A theme switch with measured contrast ratios on both backgrounds.
Not verified — and therefore not advertised
- OCR languages other than the bundled English PP-OCRv5 model.
- The Tesseract backend (code path exists; no Tesseract ships in the package).
- The Word→PDF fallback converter when Microsoft Word is absent (the Word path is what was exercised).
- Any “cloud”, “sync”, “mobile” or “web app” version — Irfan PDF is a Windows desktop application.