OCR

Copy Text Out of a Screenshot

The text you need is in an image — a screenshot someone sent you, a scanned invoice, an error message in a support ticket, a page that blocks selection. Retyping it is slow, and most OCR sites want you to upload the image first.

ShotFlow editor toolbar with the Extract Text button visible
Extract Text runs OCR on the capture you are looking at.

How to do it

  1. Get the image into ShotFlow

    Capture it, or use Open Image or Paste Image if it is already on your clipboard or disk.

  2. Click Extract Text

    The OCR engine is bundled with the extension. The first run takes a moment while it loads; after that it is quick.

  3. Review the result

    The recognised text appears ready to copy. Accuracy depends on the source: clean, high-contrast, reasonably large text reads best.

  4. Copy and use it

    Paste it wherever you needed it. The original image is untouched.

What makes it different

Runs locally, works offline

The recognition engine ships inside the extension rather than being fetched from a CDN or called as an API, so the image is never sent anywhere.

Good for the awkward cases

Error dialogs, scanned documents, screenshots pasted into tickets, and pages that deliberately prevent selecting text.

Crop first for better results

Cropping to just the text you want, before running OCR, reliably improves accuracy over running it on a full page.

Extract text, free and without an account

Nothing you capture leaves your browser unless you choose to share it.

Add to Chrome

Questions

How do I copy text from a screenshot?

Open the image in ShotFlow’s editor and click Extract Text. The recognised text appears ready to copy and paste.

Is my image uploaded to an OCR service?

No. The engine is bundled with the extension and runs in your browser, which is also why the download is larger than a typical screenshot extension.

How accurate is it?

Good on clean, high-contrast, reasonably sized text — UI screenshots, error messages, documents. It struggles with handwriting, heavy stylisation, very small text and low-contrast backgrounds.

Does it work offline?

Yes. Nothing is fetched at recognition time, so it works with no connection at all.

Can I improve the result?

Crop to just the region containing the text before running it. Less noise around the text consistently produces a better read than running OCR on a whole page.

Also in ShotFlow

Everything ShotFlow does