Images, scanned PDFs, remote desktops, videos, and some app interfaces do not expose selectable text. Copy and paste cannot help because macOS only sees the picture.
TranslateWindow captures the region you choose, recognizes the words with OCR, and places translated text back into the image where possible. You can copy the translated image to the clipboard or save it as a file.
What happens after you capture the screen
TranslateWindow first finds text blocks and their coordinates. It translates the recognized text, then draws the result back near the same blocks. This keeps labels, speech bubbles, menu items, and short paragraphs connected to the part of the image they describe.
The layout is preserved where the image leaves enough room for the translated language. A longer translation may need a smaller font or more space, so complex posters and dense tables will not always look identical to the source.
Translate a screenshot in three steps
- Open the command panel and choose screenshot translation.
- Drag around the smallest region that contains the text and its useful visual context.
- Check the translated image, then copy it or save it.
Images, PDFs, and app screens need different crops
| Source | What to capture | Useful output |
|---|---|---|
| Product image or social post | The image and its labels | A translated image ready to copy |
| Scanned PDF | One paragraph, figure, or page section | Readable text without retyping |
| Foreign app or remote desktop | The dialog, menu, or panel | Translation that keeps the controls in context |
| Video frame or game dialog | One stable frame | A translated still image |
Choose screenshot translation for one stable frame
| Situation | Better mode | Reason |
|---|---|---|
| Static image, poster, or error dialog | Screenshot translation | One capture produces a translated image |
| Subtitles or chat keep changing | Live window | The same region refreshes |
| Text is selectable | Selection | No OCR step is needed |
Clearer source text gives OCR a better chance
OCR mistakes become translation mistakes. Capture only the useful area, enlarge small text when possible, and choose a frame with good contrast. A narrow paragraph or one subtitle line is easier to read than a full dashboard with several unrelated columns.
If a name or number looks wrong, check what OCR recognized before changing the translation model. The problem may have happened before translation began.
Keep private material inside a tight region
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 send screenshots, OCR text, source text, or translations as analytics events.
Questions about screenshot translation
Can it translate a saved image? Yes. TranslateWindow can work from an image as well as a region captured from the screen.
Will the layout be identical? It keeps recognized text near its original position where possible. Dense designs and languages with longer wording may need a different fit.
Can I use the result elsewhere? Yes. Copy the translated image to the clipboard or save it as an image file.
Turn one screenshot into a translated image
Capture a focused region, keep the translated words close to the original layout, then copy or save the result.
Sources and verification
Product facts were checked against these first-party sources. Links open the current official documentation.