Image to Text (OCR)

The tool uses Tesseract, an open-source OCR (optical character recognition) engine, compiled to run inside your browser. It finds the lines of text in your image, reads them word by word and gives each word a confidence score. You get the text in an editable box, and the image shows where every line was found.

Before reading, three quick clean-ups make a real difference:

• Enlarge small images. OCR reads best when letters are around 20–40 pixels tall. A 1366 × 768 laptop screenshot is scaled up 1.76× to 2404 × 1352 pixels, so 13-pixel menu text becomes about 23 pixels tall.
• Boost contrast. The image is turned grey and stretched so the darkest ink is black and the paper is white. Dark-mode screenshots (light text on black) are flipped to dark text on white, so the engine always gets dark letters on a light page.
• Auto-rotate. Slightly tilted text is straightened. If a photo was taken sideways and almost nothing can be read, the tool tries the other three ways up and keeps the one that reads best.

Example. A 420 × 120 pixel screenshot with five lines of tiny 10-pixel grey text. Read as it is, about 90 % of the characters come out right — “last edited” turns into “ast ated” and “meeting” into “masting”. With “Enlarge small images” on, the longest side would need 2400 ÷ 420 = 5.7×, which is capped at 3×, so the engine reads a 1260 × 360 copy with 30-pixel letters — and all five lines, 40 words, come out exactly right. The first line, “File Edit View Insert Format Tools Help”, counts as 7 words and 39 characters in the box under the text.