Scanlation Busywork, Automated and Qt-ified

What it does

manga-translator-ui is a desktop and web interface built atop manga-image-translator. It runs the full scanlation pipeline—text detection, OCR, machine translation, and typesetting—then lets you tweak the results in a visual editor. It targets Japanese, Korean, and English comics, and can export layered PSDs if you still need to finish the job elsewhere.

The interesting bit

Instead of treating translation as a fire-and-forget script, the project bundles a Qt-based editor where you can deform text boxes, edit masks with a brush, and adjust fonts before final render. It also tries to keep terminology consistent across pages by extracting proper nouns automatically, which is the kind of boring detail that actually matters in long series.

Key highlights

• Ships both a Qt desktop UI and a Web UI with an admin panel
• Supports multiple OCR backends (PaddleOCR, MangaOCR, PaddleOCR-VL-1.5) and translation engines (OpenAI, Gemini, Vertex, Sakura)
• Exports editable PSDs with separate layers for original art, inpainted background, and rendered text
• Includes batch processing and workflow presets for export/import of translation text
• Windows and macOS setup scripts attempt to handle Conda, GPU drivers, and dependencies automatically

Caveats

• AMD GPU support is experimental and limited to RX 7000/9000 series; prebuilt releases exclude AMD entirely
• Docker deployment is labeled experimental and requires several mounted volumes to persist state
• Online translators require your own API keys; there is no fully offline translation engine bundled

Verdict

Worth a look if you scanlate comics and want a structured GUI rather than a pile of Python scripts. Purely casual readers probably don’t need a local pipeline this heavy.