Articles · Document Tools
Straighten a phone photo of a page without uploading it
A page on the kitchen table is not a scan. The phone adds perspective, a shadow, and whatever was behind the sheet. Uploading that photo to a “magic scanner” also uploads the letterhead, the account number, and the handwriting. You usually only need four corners you can still move, a flatten that does not invent pixels, a light clean-up that keeps the text, and a file you can print or send. The result is a better picture of the page, not a certified archive copy.
Why this is worth doing in the browser
DevNestro’s Document Scanner reads JPG, PNG, or WebP in the page, including a rear-camera capture on phones that support it. A conservative edge guess can suggest corners; you can always drag all four handles. Perspective correction maps those corners to a rectangle. Filters are Original, Clean, Grayscale, Black & White, and High Contrast, plus brightness, contrast, and rotate. Clean-up is mild on purpose so shadows do not erase ink. Keep the original until you download. Add more pages, reorder or delete them, and export one PDF with the same pdf-lib stack as JPG to PDF. Extract Text hands the flattened page to the existing local OCR tool. Nothing is uploaded. The page does not claim a perfect scan or a replacement for a flatbed.
How to use the tool
Drop or take a photo, move the four corners until they sit on the page, pick a filter you can still read, preview, then download JPG, PNG, or a multi-page PDF. Use Extract Text only after the scan looks readable. Related OCR is for text. Related JPG to PDF and Merge PDF are for files you already have.
- Shoot the page as square-on as you can; the tool corrects a lean, not a crumpled ball.
- Check the four corners before you flatten. Auto bounds are a suggestion.
- Prefer Clean or Grayscale when Black & White starts eating thin strokes.
- Export PDF when you have several pages; keep the original photos until you are sure.
Privacy
Photos stay in this browser. They are not uploaded to DevNestro and are not sent to analytics. OCR language data may download if you extract text.
Have a paper form on the table? Open Document Scanner, set the corners, and download a straightened page.