Image to text (OCR)

The reader is a bundled pattern matcher, not a neural engine: it binarizes the image, finds text lines, and matches each glyph against a built-in 5x7 bitmap font of Latin letters, digits, and common punctuation. Clean, straight, single-font text on a plain background reads well. Handwriting, skew, multiple fonts, colored or textured backgrounds, and character touching degrade to ? placeholders, and the confidence number tells you when to distrust a line. Nothing is uploaded; every pixel stays in this tab.