Select the scanned document
Choose an unlocked PDF and inspect the first-page preview. Keep the original when it contains selectable text, forms or signatures that you need to preserve.
A page placed unevenly on a scanner can leave text leaning across the sheet. Deskewing rotates the rendered page appearance, then fits it back inside the original page dimensions. This is intended for small tilt, rather than a sideways page, a curved book gutter or the perspective distortion of a camera photograph.
Choose an unlocked PDF and inspect the first-page preview. Keep the original when it contains selectable text, forms or signatures that you need to preserve.
Automatic detection estimates a small tilt for each page. A manual setting from -5 to +5 degrees adds your correction; turn off automatic detection when you want only the manual angle.
Compact, Balanced and High use different rendering and JPEG settings. Higher quality can require more browser memory and does not recover detail absent from the scan.
Check line alignment, corners and readable text in the output. The preview covers the first page; automatic detection can give a different result on another page.
Accepts one PDF up to 150 MiB with at most 80 pages. Automatic detection searches small angles between -5 and +5 degrees and can leave a page unchanged when evidence is weak.
The result rebuilds every page from a JPEG image at the original physical page size. Selectable text, links, interactive fields, accessibility structure and digital signatures are not preserved as their original objects. Rotation-to-fit may slightly reduce the page appearance.
Straighten slightly slanted text lines before making a reading copy. Inspect the result rather than assuming every scan has a reliable detected angle.
Create a straighter image-based working copy, then use Searchable OCR PDF if selectable text is required.
Turn off automatic detection and use a small manual angle when the same scanner misalignment affects the document.
Rendering, angle detection and page rebuilding happen in the browser. The scanned document is not uploaded to a deskew service. Normal website resources can still load online.
Select a question to view the answer.
No. The workflow rotates a page for small tilt; it does not dewarp perspective or curved book pages.
It estimates a correction separately for each page.
It is image-based. Use an OCR workflow afterward when you need a text layer.
Rebuilding the pages does not preserve an existing signed document as a valid original. Retain the original file.
Open another Fylvexa tool that works well with this task.