Loading
Remove scan shadows, stains, specks, and damaged margins while preserving text.
Drop PDFs or a ZIP here, or click to select
Multiple PDFs and ZIP archives are supported. Processing happens locally.
The cleanup engine is loaded when you first use this tool and cached for later runs.
Clean shadows, stains, specks, and damaged margins from scanned PDFs without uploading the document. The result has a flatter white page, clearer type, and fewer distractions when you read or print it.
Add one or more PDFs, or a ZIP containing PDFs, and choose how much page detail to render.
Start with Balanced, or open Advanced to tune the same controls exposed by ScanClean.
Clean the batch, compare every page with the draggable before-and-after view, then download one PDF or every result as a ZIP.
No. PDF.js, Pyodide, OpenCV, NumPy, and ScanClean all run in browser workers on your device. The app downloads program files, but it does not send your PDF pages to a server.
Yes. Add several PDFs or a ZIP. Pages stay sequential within each document to control memory use, and capable computers process up to two documents at the same time.
No. The cleaned PDF is rebuilt from processed page images, so existing text layers and interactive features are not carried across. Run OCR PDF afterwards if you need searchable text.
It uses ScanClean's bundled PP-OCR model as a second opinion when deciding which small marks belong to text. It runs locally through the pinned OpenCV 5 browser build and can be turned off under Advanced for faster processing.