CartoonTranslator

About the project

Built for manga fans,
by manga fans.

CartoonTranslator started as a weekend experiment to scratch a personal itch: reading untranslated manga chapters the same day they drop in Japan. It grew into a proper tool — and now it's free for everyone.

How it works

01

Upload a page

Drop in any PNG, JPEG, or WebP scan of your manga or comic. Works with raw raws, scanlation pages, and phone camera shots.

02

OCR detects the text

PaddleOCR — a state-of-the-art open-source engine — locates every speech bubble, thought box, and sound effect on the page and extracts the text with high precision.

03

AI translates in context

A large language model receives the extracted text along with positional context so it understands reading order and narrative flow, producing natural translations instead of robotic word-for-word output.

04

Rendered back onto the panel

The translated text is composited back over the original bubbles, giving you a clean, readable page — no copy-pasting required.

Tech stack

OCR enginePaddleOCR (PP-OCRv4)
TranslationLLM with manga-aware prompting
FrameworkNext.js 16 App Router
StylingTailwind CSS v4
HostingVercel Edge Network
Image processingSharp

Philosophy

Professional localisation is an art form — it takes talented translators, letterers, and editors months of work per volume. CartoonTranslator is not a replacement for that; it's a reading aid for fans who want to follow ongoing series or explore titles that may never receive an official release.

We believe tools like this should be free, fast, and privacy-respecting. Your images are never stored beyond the lifetime of a request, no account is required, and the code is open source.

If you enjoy the project, consider supporting the official English releases of the manga you love — your purchases fund the creators who make this art possible.

Ready to translate?

No sign-up required. Just drop in a page.

Try it free
Feedback