- Is my photo uploaded to a server?
- No. This tool runs entirely in your browser using WebAssembly and the Canvas API. Your photo is never sent anywhere — not to Knackpad, not to any AI cloud. This is especially important for screenshots containing passwords, card numbers, or personal documents: they stay private by design. You can even use this tool offline after the AI model files have loaded once.
- What PII patterns does the text scanner detect?
- The OCR engine reads all visible text in the image, then regular expressions flag: email addresses (user@domain.com patterns), phone numbers (common US/international formats including +1, dashes, dots, parentheses), credit and debit card numbers (13–19 digit groups matching Luhn-ish spacing), Social Security Numbers (XXX-XX-XXXX), and US street addresses (number + street name + Rd/St/Ave/Blvd/etc.). Each match gets its own black-fill box, which you can remove if it is a false positive.
- The face detector missed a face — what should I do?
- Simply drag a manual box over the missed face. The DETR object-detection model works best on clear, forward-facing faces larger than roughly 40 × 40 pixels. Very small faces in group shots, extreme side profiles, and heavy occlusion (sunglasses, masks) can reduce accuracy. For those cases, the manual draw tool is the reliable fallback — drag, position, and the pixelation applies on render.
- What is the difference between pixelate and black fill?
- Pixelation divides the region into large mosaic blocks (12 × 12 px) and fills each block with the average color of that area, making the content unreadable while preserving the visual shape. Black fill paints a solid black rectangle over the region — stronger and irrefutable, preferred for legal and compliance contexts. You can mix both styles: pixelate faces and black-fill text, or use the manual box tool to apply either style anywhere.
- Can I redact a screenshot or a scanned document?
- Yes. Screenshots (of chat messages, emails, bank statements, invoices) and scanned document photos are common use cases. For screenshots with sharp text, the Tesseract OCR step works particularly well and will catch most PII automatically. For photos of physical documents, ensure the text is reasonably in focus — Tesseract handles moderate skew but very blurry text may be missed. Always review all detected boxes before downloading.