How to Crop Margins and Clutter in PDF

Cropping margins in a PDF: what it does and what it does not

A scan with a black rim along the edge, a book photographed in spreads, a document with four-centimetre margins that has to fit a phone screen. The task is the same in every case: take the surplus off the edges. The tool handles it, but not by the method most people assume.

Cropping hides, it does not delete

Inside a PDF a page carries two rectangles. The first describes the physical size of the sheet, the second the part of it that should be displayed. Cropping edits the second and leaves the first alone.

The content beyond the visible area goes nowhere. It stays in the file in full: text comes out by copying, images come out by any program, and a proper editor brings it all back to the screen by resetting the crop.

That is not an oversight but documented behaviour: the tool explicitly marks the result as a document in which the content outside the crop was retained.

There is one important practical conclusion. If something has to be removed from a document because the recipient must not see it, cropping is categorically the wrong tool. That job belongs to redact-pdf, where the black and white styles genuinely cut the content out of the page.

The margins apply to the whole document

The second thing to know before you start: there is no page selection here. The margins you set apply to every page of the document.

For a uniform scan that is convenient: the black rim from a scanner is identical on every sheet, and one operation fixes the whole file.

For a document with mixed layout it is a constraint. A title page with wide margins, a landscape table and ordinary pages all need different crops and all get the same one.

The way around it is splitting: extract the uniform part into its own file, crop it with its own values, and assemble the document again. Slower, but predictable.

Units, and one strictness

Margins are set in points, millimetres or inches. The unit is chosen from its own list, and values are converted into the PDF's internal units automatically.

The defaults are 18 points per side, roughly 6.3 millimetres. The value was picked as "a little, but visible": it takes off the typical scanner rim and does not touch text on a normally laid out page.

There is one strictness here, and it is useful. Naming a unit that is not on the list, centimetres for instance, stops the job with a message listing the allowed options. Points are not silently substituted for centimetres, and rightly so: the difference between them is twenty-eight-fold, and such a substitution would turn a tidy crop into cutting half the page away.

How to crop the margins

1. Open crop-pdf and upload the document. 2. Set the area on the preview itself: the tool shows the page and lets you drag the frame's edges. 3. Page through the preview and make sure the frame does not touch content on the other pages. 4. Check the units if you are typing values by hand. 5. Run the job and open the result.

Point three matters more than the rest. The frame is set on one page and applied to all of them, and the typical mistake looks like this: everything is fine on page one, and on page fifteen the page number got cropped off.

How much to take off, by document type

Values are easier taken from the task than from the head. Below are figures to start from and then refine against the preview.

The last two rows mark the boundary of what cropping can do. It takes a rectangular frame off the edges and nothing else. A shadow from a book's spine runs down the middle of the spread and no crop will remove it. A skewed photograph will not be straightened by cropping: the edges come off and the text stays tilted.

Negative and oversized margins

Two cases in which the tool does not pretend everything is fine.

A negative margin value does not extend the page beyond the sheet. It is replaced with zero, and a warning listing exactly which margins were zeroed is attached to the result. So trying to "add margins" with a minus sign will not work, and will not quietly damage the document either.

Oversized margins, where nothing would be left of the page, are handled page by page. Pages where the frame fell outside the sheet are left uncropped, and the warning lists their numbers. If that turned out to be every page in the document, the job stops with a direct message saying the margins have to come down.

The second case happens more often than you would expect on documents with mixed formats: a value that suits A4 eats an A5 page whole.

Rotated pages are handled by their visible edges

A subtlety that is easy to miss and worth knowing.

You set the margins as you see the page on screen: left, right, top, bottom. On a rotated page the visible edges do not match the physical ones: turn a page ninety degrees and its physical left edge becomes the visible top.

The tool accounts for that and remaps the margins to the page's actual rotation. A value of "top 20" takes twenty points off the side you see as the top, whether the page is rotated or not.

Hence a practical recommendation: rotate first, crop second. Crop first and then rotate with rotate-pdf and the margins end up on sides other than the ones you intended.

What happens to the weight and the content

The file's weight barely changes after cropping, and now the reason is clear: the content stayed put, only the visibility frame moved.

That regularly surprises people who crop a document in order to shrink it. Shrinking belongs to compress-pdf, and it works on images rather than on margins.

Document properties and bookmarks survive a crop. The page count does not change.

The text layer survives in full as well, including the text beyond the frame. Searching the document still finds words that are no longer visible on screen, which is one more reminder that cropping is not a way of hiding anything.

When cropping is not the answer

Two tasks often phrased as cropping and solved otherwise.

Removing margins so the document takes less room on a phone screen. Cropping works here, but the result depends on the reader: some programs ignore the visible area and show the whole sheet.

Getting a picture of the page without margins. The right route then is pdf-to-jpg followed by cropping the image: there the boundary really is cut away, because an image has no notion of a hidden area.