Manga Image Translator is an open-source tool that translates text embedded in images — especially manga and comics — in one click. With over 10,000 stars on GitHub and active development into 2026, it automates the entire scanlation pipeline: detecting text regions, reading them with OCR, erasing the original lettering, and rendering the translation back into the artwork.
The project chains together deep-learning models for text detection, OCR (with strong support for Japanese, Chinese, Korean, and English), inpainting to clean the original text, and colorization-aware rendering that mimics the look of the source page. Translation itself is pluggable: you can route text through services such as Google, DeepL, or Papago, or through large language models including GPT-style APIs, DeepSeek, and Sakura for more natural, context-aware results. It runs locally as a CLI or web interface, supports batch processing of whole chapters, and can upscale pages for cleaner output. Configuration profiles let you swap detection, OCR, and translation models per language pair, and the active Discord community and frequent releases keep pace with new OCR models and LLM translators as they appear.
Free. Manga Image Translator is open source under the GPL-3.0 license. The software costs nothing; you only pay if you choose a paid translation API for the translation step.
Scanlation groups, manga and webtoon readers, localization hobbyists, and anyone who needs the text inside images translated in place — without manually editing every speech bubble.