OCR capture

Copy text from anything on your Mac screen.

Error dialogs, video captions, PDFs that block selection, and images in chat all contain text you cannot select. Shotnix runs Apple's on-device Vision framework on any region you draw and copies the result, so the text never leaves your Mac.

Menu barCapture Text, or your own shortcut

How OCR capture works in Shotnix

Choose Capture Text in the menu bar Command Center, or press the shortcut you give it in Settings → Shortcuts. Draw a selection the way you would for a normal screenshot. Shotnix recognizes the text with the Vision framework, copies it to the clipboard, and shows a toast with the result so you can confirm it read correctly. The image is kept in History too, already searchable.

If nothing is recognized, Shotnix tells you and leaves your existing clipboard untouched. A failed extraction never wipes what you had already copied.

Columns, tables, and links come through

  • Columns come out in reading order, not line by line across the page.
  • When the text holds a table, click the toast and choose Copy as Table. The rows are copied tab-separated, so they paste into a spreadsheet as cells.
  • Links and email addresses it finds show up there too, ready to open or copy.
  • Slightly tilted text still reads line by line, and Hebrew and Arabic read right to left.

Languages and accuracy

Automatic detects the language of each capture. If it guesses wrong, pick the languages in Settings → Screenshots, and choose Accurate, which reads small text best, or Fast, which answers sooner but knows fewer languages. Chinese, Japanese, Korean, Thai, and Arabic mix correctly with English. The same settings apply to Copy Text and History search.

Copy text from any screenshot after the fact

The post-capture overlay has a Copy Text action, so a normal screenshot can become text without recapturing. The same applies to captures you open from history.

Nothing is uploaded

Recognition uses Apple's Vision framework on-device. No image or text is sent anywhere, and Shotnix has no account, no server, and no analytics.

How to do it in Shotnix

  1. Choose Capture TextIn the Shotnix menu bar Command Center, under Tools, or press the shortcut you gave it in Settings → Shortcuts.
  2. Select the text regionDrag across the area that contains the text. Hold Space to move the selection instead of resizing it.
  3. Paste anywhereThe recognized text is already on your clipboard. A toast confirms what was read; if it found a table or links, click it for Copy as Table and the links.

Questions

Which languages does Shotnix OCR support?

The languages Apple's Vision text recognition supports on your version of macOS. Automatic detects each capture's language; if it guesses wrong, pick the languages in Settings → Screenshots. Chinese, Japanese, Korean, Thai, and Arabic mix correctly with English.

Can I copy a table from the screen?

Yes. When Capture Text finds a table, click the toast and choose Copy as Table. The rows are copied tab-separated, so they paste into Numbers, Excel, or Google Sheets as cells.

Does OCR work offline?

Yes. Recognition runs on-device through the Vision framework. No network connection is needed and nothing is uploaded.

Can I extract text from an existing screenshot?

Yes. Use Copy Text on the post-capture overlay, open the capture from history, or search history directly, which is OCR-indexed in the background.

How is this different from Live Text in Preview?

Live Text needs the image open in Preview, Photos, or Quick Look first. Shotnix reads text directly from a screen selection in one step, including from paused video, protected PDFs, and remote desktop sessions.

Why doesn't ⌘⇧O start text capture?

New installs of Shotnix leave ⌘⇧O alone, because it means Open in most apps. Start Capture Text from the menu bar, or give it any shortcut in Settings → Shortcuts. If you installed Shotnix before version 0.24, you keep ⌘⇧O.

Try it on your own screen.

Shotnix is free, open source, and local, with no watermark. Download the notarized app, or star the repository to follow along.

11.9 MB · Apple silicon (M1 or later) · macOS 13+ · notarized by Apple · MIT license