TypePatch

Half an hour before the meeting, someone spots last year's date in a screenshot on slide 12. TypePatch reads the words in a JPG, PNG or WebP picture and draws a box around each line it finds. Click a line and type the new wording, and the editor rebuilds the background behind the old letters first. On its own pages the tool calls itself Image Text Editor.

You give the new text a font, weight, spacing and colour, and effects such as an outline or a highlight help it match the original. Fonts for Chinese, Japanese and Korean sit next to the Latin ones. Where recognition misses a word, you draw the box by hand. The tutorials say the background repair works best on flat colours and smooth gradients, while textured photos and handwriting need closer manual work.

Recognition runs on Tesseract.js, an open-source text engine, inside the browser tab. According to the about page, images are processed on your computer, and the editor keeps working offline once the page has loaded. You download the finished picture as a PNG without signing up.