The problem we work on
Google Translate handles text. It cannot handle text inside images. Manga, e-commerce listing photos, menus, street signs, product shots, technical-doc screenshots — a huge slice of the world's content lives baked into pixels. Translating that manually + redoing layout in Photoshop takes hours per image; gluing together a generic OCR + translation API requires engineering and still loses fonts and layout. We packaged the entire pipeline into an upload-and-go tool so one person can do what used to take a team.
