Drop a PDF and every page appears as a thumbnail. Move pages earlier or later, rotate them, delete the ones you do not want, and save the result. Deleting is reversible — the page stays visible, greyed out, until you save — and there is an undo for everything else. The rebuild copies the original page objects rather than re-encoding them, so text stays selectable, vectors stay sharp and a 40MB scan comes out the same size it went in. Rotation is added to whatever the page already carries, which matters for scans: a page stored sideways turns from where it actually sits rather than snapping back to zero. The file is read and rebuilt entirely in your browser.
| Aspect | After editing |
|---|---|
| Text | Still selectable and searchable |
| Images | Not recompressed |
| Vectors | Unchanged, still sharp |
| File size | Essentially the same, minus deleted pages |
| Existing rotation | Added to, not replaced |
No. It is rendered, edited and rebuilt in your browser, and nothing is transmitted. That is the reason to use a client-side tool for contracts, records and anything else you would not paste into a stranger's server.
No. The original page objects are copied straight across rather than decoded and re-encoded, so text stays selectable, vector graphics stay sharp and images are untouched. The output is essentially the same size as the input, minus whatever you deleted.
Because many overwrite the page's rotation instead of adding to it. A page the scanner already stored at 90 degrees, rotated once more, should end at 180 — a tool that sets rather than adds puts it back at 90, so the second rotation appears to do nothing.
Yes, until you save. Deleting marks the page rather than removing it, so clicking delete again restores it, and undo steps back through everything you have done.
Not here — this works on one document. The PDF Merge tool joins files, and you can organise the result afterwards.
Your available memory, since the whole document is held in the browser. A few hundred pages is comfortable; a very large scanned archive may be slow to render thumbnails.
Pages are rendered to thumbnails with PDF.js and the document is rebuilt with pdf-lib, which copies the original page objects rather than re-encoding anything — so text stays selectable, vectors stay sharp and images are not recompressed. A rearranged 40MB scan comes out the same size it went in, because nothing was decoded and re-encoded on the way through. Rotation is added to whatever the page already carries rather than replacing it, which matters constantly with scanned documents: a page the scanner stored at 90 degrees that you turn once more should land at 180, not back at 90, and a tool that overwrites the value makes the second rotation appear to do nothing. The result is normalised into 0, 90, 180 or 270, since PDF rotation must be a multiple of 90 and pdf-lib rejects anything else — including the negatives that come from rotating anticlockwise past zero, which is where a naive implementation throws. Deletion is a flag rather than a removal, so the page keeps its thumbnail and can be restored, and the edit history keeps the last two dozen states for undo. Everything happens in your browser: the file is read, re-assembled and saved locally, which is the point when the document is a contract or a set of records.