Translation apps usually work best with real text, yet much of what appears on a screen is locked inside images, scanned PDFs, remote desktops, subtitles, and custom application interfaces.
TranslateWindow combines region capture, optical character recognition, and translation in one action. The result stays close to the source, so a visual limitation does not have to become a manual transcription task.
Where screenshot translation helps most
- Text embedded in images, diagrams, product screenshots, and social posts.
- Scanned PDFs that contain page images instead of a selectable text layer.
- Remote desktop sessions and software interfaces that do not expose text to macOS.
- A subtitle, game dialog, or video frame that needs a one-time translation.
Improve OCR before judging the translation
OCR errors become translation errors. Capture only the relevant words, use a frame with clear contrast, and enlarge small source text when possible. Avoid including icons, decorative labels, or multiple unrelated columns in the same region.
If a result looks implausible, check the recognized source before changing the translation model. A mistranslated name or number may have been misread at the OCR step.
One screenshot or a live region?
| Situation | Better mode | Reason |
|---|---|---|
| Static image or error dialog | Screenshot OCR | One capture solves the task |
| Subtitles or chat keep changing | Live window | The same region refreshes |
| Text is selectable | Selection | No OCR step is needed |
Handle sensitive screenshots carefully
A screen capture can contain more than the sentence you intend to translate. Keep the region tight and use local processing for confidential content. With a local model, recognized text can stay on your Mac; with a cloud provider, the required text is sent to that provider.
TranslateWindow does not use screenshots, OCR text, source text, or translations as analytics content. The privacy policy documents the boundary in more detail.
Translate the text your Mac cannot select
Download TranslateWindow, capture one focused region, and keep the result beside the image or app you are using.