all-in-one imager

Attempts true vector SVG export per page. If a page uses features the in-browser engine can't vectorize, that page automatically falls back to a high-res raster image embedded in valid SVG — always downloadable, never blocked.

Pulls the real text layer out of every page (not an image scan) and writes a genuine, editable .docx file — opens directly in Word, Google Docs, or LibreOffice. Complex layouts (multi-column, exact positioning) simplify to plain paragraphs, since that's what actually stays editable afterward.

Reads each word's real position on the page and reconstructs rows and columns from spacing — a genuine, working table extraction for grid-like PDFs (invoices, reports, statements). One worksheet per page. Free-flowing prose has no real "columns," so results there will just be one wide column per line — that's expected, not a bug.

Rebuilds every page as a recompressed image at the chosen quality — a real, working compression pass that runs fully in your browser. Note this flattens each page (no separate PDF-optimization library is bundled), so pages become non-selectable images — best for shrinking scanned documents or image-heavy PDFs, not text-critical ones.

Pulls the real text layer out of every page (not an image render) and compiles it into one plain-text file, page breaks included.

No files added yet.

The page grid opens on the right. Drag a page to reorder it, use its own buttons to rotate or delete it, and tick its checkbox to include it in "Extract Selected Pages" below.

Keeps headings, paragraphs, and lists. Images inside the source document aren't carried over in this version — text-heavy documents (letters, resumes, reports) convert cleanly.

Every sheet becomes its own landscape page (or pages, for long sheets), with real gridlines — the first row is bolded as a header automatically.

This one step runs on the server, not in your browser — real PDF encryption isn't something that can be done client-side.

Finds embedded photos and pulls them out as separate files. This version extracts JPEG-encoded images (the most common case — scans, camera photos) — the grid on the right opens once scanning finishes.

Detects real fillable fields (text boxes, checkboxes, dropdowns) — the field list opens on the right once scanning finishes. Works on PDFs that actually have form fields, not every PDF that merely looks like a form.

Click and drag over anything on the right to permanently remove it — the affected page is flattened to an image, so the text underneath genuinely can't be selected or copied afterward, not just visually covered.

No redaction boxes drawn yet.

Adds a real, invisible, searchable/selectable text layer under the existing scan — the page still looks exactly the same. Runs a real OCR model entirely in your browser; larger documents can take a little while, page by page.

Text: click to place, then type. Rectangle: click-drag. Image: pick a file, then click to place. Everything added is real, searchable content — not a flattened image.

No additions yet.

Shows word-level differences between the two documents' text — additions in green, removals in red — on the right once compared.

Tries a clean structural rebuild first. If the file is too damaged for that, falls back to rebuilding it entirely from page renders — this always works, but the result's text is no longer selectable/searchable (you'll be told which happened).

Each page becomes one full-slide image — visually exact, but text inside each slide isn't independently editable (the whole slide is one picture).

Converts pasted/uploaded HTML content — not a live URL (browser security blocks reading another site's page from client-side code, which would also break the "nothing leaves your browser" guarantee every other tool here keeps).