Chat Screenshot Cleaner

Choose screenshots or use Try sample screenshots to learn the controls. New images are scanned for text automatically, but you decide what gets covered. Stop pauses processing; use Draw a cover or Add adjustable cover when you already know what needs hiding.

Small lettering, compression, dark themes and coloured bubbles can confuse text recognition. Try the original full-size screenshot, change Screenshot appearance and use Scan text again. The reader uses English data; cover missed text manually rather than assuming an empty result means nothing private is present.

Enter the name as it appears in the detected text under Name, username or other text. Review the matches, add an Optional anonymous label such as Person A and cover them. Matching ignores letter case but depends on recognised whole words, so spelling errors and unread words can be missed.

Draw a tight cover around the avatar, select it and use Find this avatar in all screenshots. It compares similar pictures at a similar size, not people’s identities. Check every added cover: different crops, themes, sizes or compression can cause misses and false matches.

This cleaner uses solid, opaque covers. Blur softens detail and pixelation groups it into blocks; either can leave text or a face recognisable and should not be treated as dependable redaction. A solid cover replaces the pixels inside its area in the flattened export. Extend it beyond the detail’s edges and check the saved image for anything left visible.

The PNG contains the edited pixels, not removable cover layers or the original screenshot underneath. Original files and recognised text are not attached. That does not make the whole conversation anonymous: message wording, timestamps, replies and uncovered details can still identify people. Source metadata is not copied, but this is not a forensic guarantee about every possible identifying trace.

It records that you checked that screenshot; it is not an automatic privacy certification. Look at names inside messages, avatars, notification bars, contact suggestions and the edges of covers. Editing a screenshot requires a fresh review before download. Face and contact detection are suggestions, not a complete check.

Put screenshots in reading order, trim repeated headers and use Suggest overlap trims if sections overlap. Review every join for duplicated or missing messages. One long PNG shrinks images to a common width without enlarging them; Separate PNG images in a ZIP keeps each image’s width.

A batch allows eight images, 15 MB per file, 12 million pixels per image and 32 million input pixels in total. Input images and the joined result are limited to 16,000 pixels per side; the joined result also has a 24-million-pixel limit. Try a smaller Joined image width, separate PNGs or a smaller batch.

Text recognition, face detection, matching and image export run in your browser tab. They do not send your screenshots or recognised words to a processing service. Detection files download from ToolsNow when needed, and the page can make other resource requests. Local processing does not mean the page is network-free.