Reorder PDF Pages, Delete & Rotate

About Reorder PDF Pages, Delete & Rotate

Page order in a PDF is nothing more than the order of the Kids array in the page tree. Reordering rewrites that array and leaves every byte of page content alone. Alongside reordering, this tool gathers the other page-level jobs into tabs: delete, rotate, duplicate, insert, interleave duplex scans, mirror, remove blank pages and remove duplicate pages. All of them decide which pages stay and how they sit.

Because the original file is kept as the base and only its page list is rebuilt, bookmarks, forms and page labels come through. Bookmarks and form fields that pointed at deleted pages are cleaned up with them, and images and fonts used only by deleted pages are dropped from the file.

Rotation turns the view, not the content

Every PDF page can carry a /Rotate value. The specification says it must be a multiple of 90 and tells viewers and printers to turn the page that many degrees clockwise. This tool rotates by adding to that value: a page already at /Rotate 90 turned right once more becomes 180, and the coordinates in its content stream never change. You can rotate as often as you like with no quality loss, and rotating back gives you the original exactly.

The catch is that software working with coordinates — text extraction, some conversion and imposition tools — sees the page before /Rotate is applied. Straighten a sideways scan with rotation and a few pre-OCR tools may still treat it as landscape; in that case the page needs to be redrawn, as the auto-orient mode of the scan-cleanup tool does. Mirroring works the other way: it inserts a reflection matrix (a cm operator) in front of the content stream, so the drawing itself is flipped. That’s what you want for iron-on transfer paper, where the print must come out reversed.

How blank-page removal decides a page is blank

Blank-page removal renders each page 800 pixels wide, ignores a 5% border and counts a pixel as ink only if its brightness is below 160 — the same test as the blank-page mode in Split PDF. The default threshold of 0.05% works out to 335 pixels on a US Letter page (800×1035 px, 720×933 px inside the border, about 672,000 pixels) and 366 on A4. A truly empty page exported from Word has zero ink pixels and is caught every time.

Recycled or yellowed paper scans with a background around RGB 235, far brighter than 160, so it doesn't count as ink. Punch holes and the dark strip at the edge of the scanner glass fall inside the ignored border. On the other hand, pencil notes or faint stamps lighter than 160 aren't counted, so such a page can come up as a blank candidate, and a staple shadow or dark stain that reaches into the body still counts as ink. Detect only selects candidates: check the thumbnails and untick any page you want to keep.

Finding duplicate pages from a double feed

When a document feeder grabs two sheets or someone re-feeds a page, near-identical pages end up next to each other. This mode shrinks each page to a 32×32 grayscale grid and compares in two stages. First it checks a 64-bit difference hash (dHash), which records whether each cell is brighter than its neighbour; at normal sensitivity a pair may differ in at most 10 of 64 bits. Pairs that pass are then scored on the average brightness difference across the grid, and need at least 97% similarity. Contrast is stretched first so that two scans of the same sheet at slightly different exposure still match.

If both pages have a text layer, you can require identical text too. Turn that on to protect pages that look alike but aren’t — monthly statements that differ only in the date and amounts. The tool only proposes candidates; it deletes just the pairs you confirm. Blank pages all look the same, so they’re left out of the comparison by default.

Interleave: duplex documents from a simplex scanner

With a one-sided scanner you scan the fronts, flip the stack and scan the backs. The flipped stack feeds from the last sheet’s back, so the backs file comes out in reverse. Load the fronts file, then the backs file, tick “reverse backs”, and the result reads front 1, back 1, front 2, back 2 and so on. The tool proceeds only if there are as many fronts as backs or exactly one more, which allows for a final sheet with a blank back that you didn’t scan.

Duplicating and inserting barely grow the file

Duplicating a page 50 times doesn’t make the file 50 times bigger. Each copy is a thin page dictionary pointing at the same content stream, fonts and images by object number. Copies can be grouped (1,1,2,2) or collated (1,2,1,2), up to 100 per page and 5,000 pages in total. Form fields and pop-up notes can’t be safely shared between pages, so the copies leave them out.

Insert adds blank pages, another PDF or a JPG/PNG image at the start, the end, before or after page N, or after every N pages. An image is embedded once and shared, so adding the same divider image every four pages for duplex printing only costs one image’s worth of space.

Worked examples

How to use it

1. Load a PDF and pick a tab: order, delete, rotate, duplicate, insert, interleave, mirror, blank or duplicate removal.
2. Drag or select pages in the thumbnail grid and set the mode’s options.
3. Check the resulting page count and download the PDF.

Frequently asked questions