LIBRE MANGA
TRANSLATOR

Raw scan in.
Translated page out.

LMT erases original text with a fitted text-shaped mask and repaints your language inside the same bubble — screentone and artwork survive untouched. Keep scrolling to wipe between the raw page and the result.

• Fast / Quality ladder — planar fill → bilateral denoise → pure-JS Telea, decline-gated, or LaMa neural redraw (opt-in) with automatic Fast ladder fallback and missing-weights guard
• Every rung scored by one quality check — a region nothing can clean is left pixel-identical and flagged declined
• Minimal ghost rectangles, no flat patches, no halos

Three modes.
One pipeline.

Libre — you choose where your data goes. Default to fully on-device Local Mode, take the cloud when your GPU is weak, or keep total control with your own server. Toggle back to the original image any time.

• 8 runner presets (Ollama, LM Studio, JAN, Unsloth, llama.cpp, KoboldCpp, LocalAI, Cherry Studio) + collapsible setup guides
• Dual schema support: OpenAI json_object and LM Studio experimental payloads with Test Connection verification
• Model Storage lists every weight with size — per-model delete, full cache clear, and inline downloads with pause/resume

Heavy inference,
off the main thread.

An RT-DETR bubble detector runs in a dedicated offscreen document — the page you're reading never drops a frame. Sliding-window tiling with IoU deduplication seamlessly handles 10,000+ px webtoons without memory overflow. Comic Text Detector with pixel-mask seeding is selectable in Vision › Detection.

• Detection tiling — 1500px sliding windows with IoU deduplication and aspect-ratio preservation for ultra-tall manhwa
• Language gate & OCR tracking — 3.7 MB script-ID model holds back sound effects; unified slotIndex correlates batch positions

The box editor
is honest.

Detected boxes are numbered in manga reading order. Drag a missed bubble into place, resize the ones the detector got wrong, undo and redo freely — before a single token is spent on translation.

• Right-to-left, top-to-bottom numbering with dedicated move handles and panel-aware band grouping
• Skip the editor entirely with Skip bbox refining, or batch whole chapters hands-free with Auto-translate
• Overlay isolation + natural-vs-rendered scaling fix — editor keys never fire the host reader

Read the OCR.
Fix the line.

Open the Typeset Mode to see the extracted text and the translation side by side, per bubble. Tweak a line, hit Apply — it re-renders on the canvas from the cached inpainted base, without re-running inpainting.

• Gate-held regions keep their recognized text and show a one-click Translate anyway
• Toggle original vs translated at any moment to compare
• Multi-layer editor — switch Raw / Cleaned / Result, paint inpaint patches or restore art, style per bubble

Your site,
your rule.

LMT matches the current URL against community adapters to pull series name, chapter ID and page index. If your site isn't covered, the AI rule generator drafts one from your active model — or write it by hand in about three minutes.

• Drop a file in src/lib/adapters/ — auto-bundled at build, no registry edits
• Floating on-page sidebar: switch mode mid-chapter without leaving the page
• Anti-hotlink bypass + magic-byte decode — JPEG, PNG, WebP, GIF, AVIF

Three steps
to your first page.

No cloud subscription, no account, no credit card. In Local Mode the whole pipeline works offline after the initial model download. Same-device API Mode (Ollama, LM Studio, JAN on localhost) is offline-capable too.

• Right-to-left box sorting is already done for you
• Results cache per page — re-opening a translated chapter is instant
• Self-host on the same device — 8 runner presets including Ollama 127.0.0.1:11434, LM Studio 127.0.0.1:1234, JAN 127.0.0.1:1337, no internet needed

Built in the open.
Shipped honestly.

All core translation functionality is fully released and production-ready across Chrome and Firefox. Development focuses on hardening OCR reliability, reader ergonomics, and fast bug resolution.

• Continual OCR config tuning for all OCR Engines
• OCR reliability on Korean manhwa / webtoon — Pororo handles Hangul offline but still trails dedicated Python engines on stylized art text

Built on open work.
Keep it Libre.

LMT is AGPL-3.0 — copyleft keeps this fork open: anyone may redistribute or build on it, but every distributed or network-hosted derivative must keep its complete source available under the same terms.

• ComicTL by Ketut Shridhara — detection pipeline, PaddleOCR, WebLLM translation, bbox editor, AI context, adapters and Gemini path
• Bear Ladder inpainting concepts (GPL-3.0, KoMiQ) — fitted masks, ladder + gate, denoise — independent TS rewrite, no copied code
• Weights: RT-DETR Apache-2.0 · ComicTextDetector GPL-3.0 · OSD Apache-2.0 · Paddle/PP-OCRv6/Manga-OCR/Pororo Apache-2.0 · LaMa MIT

100% on-device by default · privacy.md ↗