Capture2Text can do more than copy text off your screen — it can translate it on the fly. This guide explains exactly how the built-in translate feature works, how to turn it on, and the two facts every user should know upfront: translation is powered by Google Translate, and it requires an internet connection. (Everything else in Capture2Text — capture, OCR, clipboard — works fully offline.)
If you have not installed the tool yet, grab it on the download page — Capture2Text v4.6.3 (2022-03-19) is free, open source (GPLv3), portable, and Windows-only.
How to enable the translate feature
- Right-click the Capture2Text system-tray icon and open Settings.
- Go to the Translate tab and check Enable Translation.
- Set your target language (the language you want to read).
- Choose what happens with the result: replace the clipboard text, show a popup, or both.
- Test it: press Win+Q, capture some foreign-language text, and check the output.
The source language is auto-detected by Google Translate, but remember the OCR step comes first — the OCR language must match the text on screen. Capturing Japanese with English OCR data selected produces garbage that no translator can fix. Install the right data via how to add OCR languages; Japanese users should also read Japanese OCR and managing languages.
It is Google Translate — here is what that means
Capture2Text does not ship its own translation engine. Captured text is sent to Google Translate, and the translated result comes back. Practically, this means:
- You must be online. No connection, no translation. This is the one Capture2Text feature that is not offline-capable — a deliberate qualification to any “100% offline” claim about the tool.
- Translation quality equals Google Translate quality for your language pair — good for getting the gist of menus and dialogue, rough for slang, jokes, and literary text.
- Your captured text is sent to Google. Do not translate passwords, private documents, or anything sensitive. For private material, capture offline and translate locally by other means.
- Very long captures may hit request limits; translate in chunks if output stalls.
For longer documents, translate in sections rather than capturing an entire page at once. Smaller captures OCR more accurately, translate faster, and are easier to proofread — a mistranslated sentence is much easier to spot (and fix) in a three-line capture than in a wall of text. If you translate the same kind of material regularly, keep a small glossary of how key terms come out; machine translation is consistent enough that recurring terms usually translate the same way each time, which makes proofreading faster over time.
Common workflows
- Manga readers: Bubble OCR a speech bubble, get a rough English gloss in seconds — the full setup is in the manga reader’s guide.
- Gamers: capture dialogue boxes during play — see real-time game translation.
- Students: capture textbook passages and unknown phrases, then turn them into flashcards — learn Japanese with Capture2Text shows the method.
- Office documents: screenshots of foreign-language slides or PDFs — the screenshot-to-text workflow covers proofreading and export.
DeepL vs Google Translate: the honest comparison
Many users ask whether Capture2Text can use DeepL instead. The honest answer: Capture2Text has no DeepL integration — translation goes through Google Translate only, and there is no setting to change the provider.
That said, DeepL genuinely produces better translations than Google for several language pairs (Japanese→English and German→English are the commonly cited examples). If translation quality matters more than speed, the practical workaround is: capture with Capture2Text, then paste the clipboard text into DeepL’s web translator yourself. It is one extra step, but the output is often noticeably better.
To be completely clear: we are not affiliated with, partnered with, or endorsed by DeepL (or Google) — this page mentions DeepL only as an honest user option, with no commission links of any kind.
Troubleshooting translation
- No translation appears: check your internet connection first, then confirm translation is enabled in settings.
- Garbled output: the OCR language probably does not match the source text — fix the OCR side, not the translation side.
- Wrong target language: re-check the target language setting; it is easy to leave it on a previous language.
- Nothing happens on capture: verify the capture hotkey itself works (see hotkey setup) before blaming translation.
Where to go next
- Manga reader’s guide · Game translation · Learn Japanese
- Add OCR languages · Managing languages
- Capture2Text vs QTranslate — do you need a dedicated translation tool too?
- All guides · Download Capture2Text
Frequently asked questions
How do I turn on translation in Capture2Text?
Open Settings from the system-tray icon, go to the Translate tab, enable translation, and set your target language. Then capture text with Win+Q as usual.
Does Capture2Text translation work offline?
No. Translation is powered by Google Translate and requires an internet connection. Capture and OCR themselves work fully offline — translation is the single online-dependent feature.
Can Capture2Text use DeepL instead of Google Translate?
No. Capture2Text has no DeepL integration and no provider setting. The workaround is to capture text with Capture2Text and paste it into DeepL’s web translator manually. We are not affiliated with DeepL.
Why is my translation gibberish?
Almost always an OCR problem, not a translation problem: the OCR language data does not match the text on screen. Install the correct language data and re-capture.
Is it safe to translate sensitive documents?
No. Captured text is sent to Google Translate over the internet. Never translate passwords, contracts, medical records, or anything private with this feature.
Which languages can it translate?
Whatever Google Translate supports — the OCR side must have the matching Tesseract language data installed (English, French, German, Japanese, Korean, Russian, and Spanish ship with Capture2Text; more can be added).