Versely
    Guides

    Copy text from a screenshot on any device

    Built-in ways to copy text from a screenshot on iPhone, Android, Windows and Mac, when a browser OCR tool wins, and where on-device OCR still struggles.

    Versely Team••7 min read

    You have a screenshot with text you need: an error message, a quote, an address, a code from a slide. Retyping it is slow and error-prone. Every major operating system can now read text from an image without an extra app. This is where each built-in option lives, when a browser tool does the job better, and where all of them still fail.

    iPhone and iPad: Live Text

    Apple's Live Text reads text in photos and screenshots in the Photos app. Per Apple's support page, you need an iPhone XS, iPhone XR or later with iOS 15 or later.

    1. Open the screenshot in Photos.
    2. Touch and hold a word, then drag the grab points to set the selection.
    3. Tap Copy, or tap Select All to take everything.

    If nothing happens, check that Live Text is turned on under Settings, General, Language & Region. Live Text is available in some languages and regions, not all, and Apple keeps that list on a separate page.

    Where it wins: one screenshot, a few lines, and you are already on the phone. Nothing is faster.

    Android: Google Lens

    On Android, the usual route is Google Lens from the photo viewer. In Google Photos, open the screenshot, tap Lens, and then, as Google's help page puts it, "tap and select the words you'd like to copy". Phone makers ship their own gallery apps, so the button can sit in a different place. Look for the Lens icon.

    Where it wins: same as Live Text. It is quick and built in, and good on clear text.

    Windows: Snipping Tool and PowerToys

    Windows 11's Snipping Tool has OCR built in. After you capture a snip, select the Text actions button. You can select and copy some of the text, use Copy all text, or use Quick redact to hide email addresses and phone numbers. Microsoft says all text recognition runs locally on your device.

    If you want text straight to the clipboard with no image at all, Microsoft's free PowerToys includes Text Extractor. Press Win+Shift+T, drag a box over any part of the screen, and the text in it is copied. Two caveats from Microsoft's own page: proofread the output, and it can only read languages whose Windows OCR language pack is installed.

    Where it wins: text on your screen right now, in any app, including a paused video.

    Mac: Live Text in Preview

    On a Mac, open the screenshot in Preview. Hold the pointer over the text, drag to select, then Control-click and choose Copy Text. If you cannot select text, choose Tools, then Text Selection. That is from Apple's Preview guide.

    Where it wins: a Mac screenshot you already have open. There is no install and no upload.

    When a browser tool does it better

    The built-in options are designed for grabbing a few words from one image. They are clumsy for other jobs:

    • A batch. Ten screenshots of a thread, a set of slides or a stack of receipts, and you want all the text in one file.
    • A machine you do not control. A work laptop that blocks installs, a shared computer, a Linux desktop, or an older Windows version without the new Snipping Tool.
    • Editing before copying. You want to correct the lines, then copy or save them, in one place.
    • Seeing what was read. You want boxes showing where each line was found, so you can spot a skipped line.

    That is what the free image to text tool is for. You can drop, paste (Cmd or Ctrl+V) or pick up to 10 images at a time, JPG, PNG or WebP, up to 40 MB each. A vision model, Microsoft's Florence-2, reads each one in your browser. It uses WebGPU when the browser has it and WebAssembly when it does not. Each image gets an editable text box. You can copy one image, copy the whole batch, or download a .txt. Turn on text regions to see where a second pass found each line.

    Nothing is uploaded. The one network request is a first-run download of the model, about 360 MB on WebGPU or 230 MB on the WebAssembly fallback. After that it loads from the browser cache.

    Be honest about small text on dark screenshots

    This is the part most OCR pages skip. On-device OCR is good at clear printed type: slides, documents, simple receipts, big interface labels. It is much weaker on:

    • Small text on dark interfaces. Grey 11-point text on a near-black background in dark mode is hard to read for any OCR. It is especially hard for a small model running in a browser tab. You get dropped words, merged lines, or in bad cases a phrase that repeats. The tool carries a guard against that repetition, added because small light-on-dark interface text made the model loop.
    • Tall phone screenshots. The image to text tool scales each image to 1,024 pixels on its long edge before reading. A phone screenshot 2,500 pixels tall is shrunk to about 40 percent of its size. Text that was small becomes tiny.
    • Long pages. The model stops after a fixed amount of output. A full page of small type can be cut off at the end, and the tool tells you when that happens.
    • Handwriting, skewed photos and dense tables. Words get dropped and columns get merged.

    The fixes are practical:

    1. Crop to the part you need before you drop it in. A cropped message fills the 1,024 pixels. A full screen does not.
    2. Zoom in before you take the screenshot. Bigger text on the screen means bigger text in the image.
    3. Switch the app to light mode and take the screenshot again, if you can. Dark text on a light background is easier to read.
    4. Split long pages into two or three screenshots.
    5. Read the output before you paste it anywhere important. A wrong digit in a phone number or an order reference looks just like a right one.

    If a built-in tool and the browser tool both struggle with a screenshot, the text is probably too small in the image. Retake the screenshot before trying a third tool.

    Which to use

    • One screenshot, on your phone: Live Text on iPhone, Google Lens on Android.
    • Text on your Windows screen now: Snipping Tool Text actions, or PowerToys Text Extractor.
    • A screenshot open on a Mac: Preview.
    • A batch, a locked-down machine, or text you want to edit and save as a file: image to text in the browser.

    Related jobs that are not OCR

    A few look-alike tasks need a different tool:

    • A PDF that already has selectable text does not need OCR. PDF to text pulls the real text layer out, which is exact. A scanned PDF has no text layer; export the page as an image and use image to text.
    • Speech in a video is transcription, not OCR. Use video to text.
    • A description of the picture, for an image or video model, is image to prompt. It uses the same Florence-2 model.
    • Screenshots you are about to post often contain names, emails and order numbers. The image redactor reads the text with the same model and flags those for you. Our checklist of what to redact before you post a screenshot covers the rest.

    Versely's main product is AI video and image generation, on the web and in the mobile app. It is useful when the text you copied is the start of a script or a caption for a new clip. The OCR itself is a free job in your browser, and it should stay that way.