To translate text in an image, each line of text inside the picture is replaced with its translation and redrawn in the same typeface, weight, colour and position. EditImageText first finds every line of text, then asks the image model to rewrite the whole picture in the target language, and finally copies only the text areas back into your original file — so backgrounds, photos and illustrations stay identical.
That is the difference from camera translators such as Google Lens or Apple Live Text, which overlay a translation so you can read a sign. Here the result is a production asset you can publish: a localised store screenshot, a menu photo in a second language, a banner variant for another market.