How to extract text from an image on Mac

The steps

1. Open the best image

   Use the original image in a viewer and zoom until the text is clearly visible.

2. Select the text region

   Choose Capture Text in LensOCR and drag around one coherent passage.

3. Paste the words

   Click into your document and press Command-V.

4. Compare against the image

   Check spelling, punctuation, and important numbers before using the extracted text.

Choose a clear image rather than a small preview

A picture of a document, a screenshot, or a photographed notice can contain useful words without an editable text layer. Open the image itself in Preview or another viewer instead of working from a tiny thumbnail. If someone sent several versions, start with the largest original that keeps the letters sharp.

Zoom until the text is comfortable to read, then look at a few individual characters. If those shapes are already blurred or cut off, the extracted text will need more checking. When you can take another photo, frame the page straight on, keep it in focus, and avoid glare over the passage you need.

Prepare the image on your Mac screen

Place the viewer window where it is easy to see and close any panels covering the image. Rotate a sideways picture in the viewer so the lines are upright. For a long image, zoom to a readable section and work through it in parts rather than shrinking the whole picture until its text becomes tiny.

You do not need to turn the image into a different file format for this screen-based workflow. LensOCR recognizes the pixels visible in the selected area. An image displayed in a web page or chat window can therefore be captured too, provided the words are visible and the app has macOS Screen Recording permission.

Capture one passage from the image

Choose Capture Text mode from LensOCR's menu bar menu. Start the capture with the menu action or your configured capture shortcut, then drag around the desired words. Include the complete passage and leave a little room around the characters. Press Escape if you need to cancel, move the image, or choose a better region.

A receipt or flyer may have separate blocks of text, each with its own reading order. Capture those blocks separately when their arrangement would make one large extract confusing. For example, copy a product description apart from a nearby price list. Paste each result before starting another capture, since the clipboard updates with the next recognized passage.

Turn recognized words into an editable document

Click into Notes, Pages, an email draft, or the text editor you use and press Command-V. The pasted words are editable, so you can fix a recognition error or reorganize the passage. Keep the original image beside your document until you have checked it. For a longer extract, compare one paragraph at a time.

The image's visual design does not become a matching document layout through text recognition. Recreate headings, columns, and emphasis in your editor if they are needed. For receipts, add your own labels to amounts so a subtotal is not confused with the final total. Preserve the original image if its arrangement carries meaning.

Adjust recognition for language and unusual words

LensOCR can detect the recognition language automatically. Its recognition settings also let you select from the languages supported by the current macOS Vision framework. Manual selection can be useful when an image contains a short phrase or specialized material. The available list depends on macOS, so use the choices shown in your installed app.

Try language correction for ordinary prose, and add recurring technical terms to the custom dictionary when they are often misread. Review names and codes carefully because a plausible word is not always the right word. Ignore line breaks can join recognized lines into a paragraph; leave it disabled when a list or address needs distinct lines.

Review text that will be reused elsewhere

Before sending or publishing the extract, check the details that matter in context. In a meeting note, that may be a person's name. In a photographed label, it may be a model number or a unit of measurement. For a quotation, compare every sentence with the source and add the attribution yourself. A clear capture reduces cleanup, but proofreading remains part of the workflow.

Questions about this workflow