ToolPebColour Document Scanner & OCR
Colour-preserving cleanup, shadow reduction, editable OCR text and PDF/PNG downloads.
Clean up document photographs while keeping their colours. Read printed text with OCR and export your scan.
Camera: on a phone, choose your camera from the file picker. Up to 30 pages; 50 MB per file. Choose standard or higher scan detail below. Batches also have a memory budget; use smaller batches on phones.
Changing detail affects new imports. Originals here are the imported-resolution pages; keep the original files separately.
Add a document to begin. No high-contrast black-and-white filter is applied.
Brightness is corrected with shared RGB gain to protect hue. Strong cleanup may alter faint print or paper texture; compare both previews.
Document edges & perspective
Suggest the corners of a bright page on a darker background, or enter them manually. Inspect the outline before applying. This flattens a planar page; it does not reconstruct curved folds.
Crop page margins
Download your scan
Read the printed text
Run OCR after your final page edits. Printed text works best; handwriting and damaged text may be inaccurate.
Text corrections are included in searchable PDFs. Unedited words keep their OCR positions; corrected pages use an approximate invisible reading-order layer, while the visible scan stays unchanged. English and Devanagari fonts are bundled. Review text order and recognition before sharing.
What cleanup can and cannot restore
Filters reduce uneven lighting, broad shadows and some crease shading. They keep the source image separate and do not invent missing letters. Physical folds, severe wrinkles, warped geometry, tears, stains covering text, glare and blurred or hidden information cannot be reliably removed by these filters. Flatten the page and retake the photograph when possible. Rotation, crop, suggested bright-page boundaries and manual four-corner perspective correction are included. Edge suggestions require review. Curved-page dewarping and restoration of hidden information are not implemented.
OCR is local Tesseract.js, with English, Hindi and Marathi language models bundled. PDF import/export and OCR require a local or hosted HTTP website; use START_WEBSITE.cmd. No image or text is uploaded by this scanner.