Japanese OCR with Capture2Text: Vertical Text & Bubble OCR

Japanese Works Out of the Box

Japanese (jpn) is one of Capture2Text’s seven bundled languages — the language data comes from the shared tessdata repository — no dictionary download needed, unlike Chinese. Right-click the tray icon → OCR Language → Japanese, and you are ready. Two Japanese-specific features make Capture2Text unusually good for learners and manga readers: automatic furigana stripping and Bubble OCR.

Text Direction: Win+O Is Your Best Friend

Japanese runs both horizontally (websites, subtitles) and vertically (manga, novels, signs). With Japanese active, press Win+O to cycle horizontal / vertical / auto. On auto, Capture2Text picks horizontal when the capture box is wider than twice its height, otherwise vertical. The direction setting also affects furigana stripping, so set it deliberately rather than leaving it to chance.

Furigana Stripping

When Japanese is the active language, Capture2Text automatically attempts to strip furigana — the tiny reading kana printed above kanji — from the OCR output. You get clean 日本語 instead of readings tangled into the text. It is heuristic, not perfect, but it saves enormous cleanup time for learners mining sentences.

Bubble OCR: The Manga Feature (Win+S)

Bubble OCR Capture is a real, documented Capture2Text feature built for comics:

  1. Position the mouse in the empty part of the speech/thought bubble — not on the text.
  2. Press Win+S.
  3. Capture2Text detects the enclosed bubble and OCRs the text inside, straight to your clipboard.

The bubble must be completely enclosed. For standard rectangular dialogue or narration boxes, the regular Win+Q box capture or Win+E line capture is faster — see the hotkey reference.

From Capture to Study: JGlossator

The official Capture2Text documentation itself recommends JGlossator (Windows, free and open source) for Japanese learners: paste or pipe your OCR’d words in and get de-inflected expressions, readings, audio pronunciation, example sentences, pitch accent, kanji info and grammar analysis, backed by EDICT/EPWING dictionaries. The power-user setup: Settings → Output → Call Executable with ${capture} to send every capture straight into your lookup workflow — pattern explained in our automation guide. If you would rather look words up inside your browser, Yomitan is the other popular option — see Capture2Text vs Yomitan: complementary, not rivals.

Quick Workflow for Manga / VN Readers

  1. Set Japanese as a Win+1 quick-access language (language guide).
  2. Win+O → vertical for manga pages.
  3. Win+S on bubbles, Win+E on narration lines.
  4. Optional: enable translation (Settings → Translate, needs internet) or pipe to JGlossator.

Reading Chinese too? See the Chinese OCR guide. New to the app? Start with installation or browse all guides.

Frequently Asked Questions

Does Capture2Text support Japanese OCR?

Yes — Japanese is one of the seven languages bundled by default, so no download is needed. Select Japanese from the tray icon’s OCR Language menu and capture with Win+Q. Capture2Text even strips furigana readings automatically.

How do I OCR vertical Japanese text?

With Japanese active, press Win+O to cycle text direction: horizontal / vertical / auto. Auto uses vertical when the capture box is taller than twice its width — perfect for manga dialogue and novel pages.

What is Bubble OCR and how do I use it for manga?

Bubble OCR (Win+S) captures text inside comic speech/thought bubbles: place the mouse in the empty part of the bubble (not on the text) and press Win+S. The bubble must be fully enclosed. It is ideal for manga raws.

Does Capture2Text remove furigana?

Yes — when Japanese is the active OCR language, Capture2Text attempts to automatically strip furigana (the small reading kana above kanji) from the output. The text-direction setting (Win+O) affects how this works.

Can I look up the Japanese words I OCR?

The official documentation recommends JGlossator (Windows, open source) for looking up OCR’d Japanese words with de-inflected expressions, readings, audio, example sentences and pitch accent. You can also pipe captures to it via Settings → Output → Call Executable (${capture}).

Can Capture2Text translate Japanese to English?

Yes via Settings → Translate tab (Google Translate, requires internet). Many learners instead pipe ${capture} to a dictionary tool like JGlossator for deeper study — see our AutoHotkey automation guide for the Call Executable pattern.