Paper rarely feeds into a scanner perfectly straight, and phone photos of pages are often a few degrees off. The tilt is easy to ignore while reading but obvious in print, and it measurably hurts text recognition, because OCR expects lines of text to run level. Deskew PDF measures how far each page leans and turns it back to straight, or lets you set the angle yourself, page by page.
How to use Deskew PDF
- Add the scanned PDF.
- Keep Automatic to have every page measured, or choose Manual to set each page's angle while you watch the preview.
- Select Straighten Pages and check a few pages of the result.
- Run OCR PDF next if you need searchable text.
How pages are straightened
- Measured page by page: Each page is drawn at low resolution and turned in small steps until its lines of text run level. The angle that lines them up best is the page's lean, found to a tenth of a degree.
- Turned in place: The page's content is then turned by that angle around its centre. Nothing is redrawn as a new picture, so the scan keeps its resolution and any text, links or form fields on the page come along.
- Only the pages that need it: Pages that already run level are left untouched. Tick Only straighten strongly skewed pages to leave slight leans of under two degrees alone as well.
Automatic or manual
- Automatic: The measured angle is applied to every page that leans. This suits scans of printed text.
- Manual: For pages the measurement can't judge, such as photos, drawings or pages with little text, set the angle yourself in steps of 1, 0.5 or 0.1 degrees. The preview turns as you go, and the arrows above it move to the next page.
Straighten before OCR
- Why tilt breaks recognition: Tesseract, the engine behind OCR PDF, reads text most reliably when lines are level. Even a couple of degrees of tilt can break words across lines and increase misreads.
- One of the best improvements: Deskewing first is one of the most effective ways to improve OCR on phone scans and fed paper.
- Where to read more: The searchable PDF guide and the scan quality guide explain how tilt, contrast and resolution affect recognition.
What to expect in the output
- Quality and text are kept: Because the page content is turned rather than redrawn, the result is as sharp as the scan, and an existing text layer stays searchable.
- Corners: Turning a page leaves thin white wedges at its corners where the scan no longer reaches. Crop PDF trims them if they show.
- How far it turns: Automatic straightening corrects leans of up to 15 degrees; by hand you can turn a page up to 45 degrees either way.
- Sideways or upside-down pages: A page that's sideways or upside down needs Rotate PDF.
When people use Deskew PDF
Preparing scans for OCR
Straighten every page before text recognition so names, numbers and dates are read more accurately.
Cleaning up fed paper
Fix the gradual lean that sheet feeders introduce across a long batch scan of invoices or letters.
Neater phone scans
Level pages photographed on a desk so the document looks intentional when printed or shared.
Limitations to know
Deskew PDF corrects rotation of up to 15 degrees automatically and up to 45 degrees by hand. It doesn't fix upside-down pages or the perspective distortion of photos taken at an angle, and automatic measurement works best on pages with lines of text.
Deskew PDF: common questions
Will deskewing remove the text layer from my PDF?
No. The page is turned, not redrawn, so text that was searchable before stays searchable.
Why did some pages not change?
They already ran level, or leaned less than a tenth of a degree. With Only straighten strongly skewed pages ticked, pages leaning less than two degrees are left as they are too.
Can it fix pages that are upside down?
No. Deskewing corrects slight tilts. Use Rotate PDF for pages turned 90 or 180 degrees.
Does it fix photos of pages taken at an angle?
It corrects rotation, not perspective. A page photographed from the side, so that it looks like a trapezoid, stays distorted. Retaking the photo straight on gives the best result.