Snow ShotHow to Copy Text from a Screenshot on Windows 11 (Step by Step)
Windows 11 has no built-in way to copy text out of a picture. If someone sends you a screenshot of an error message, a receipt or a slide, you retype it — unless you use OCR. This guide walks through copying text from a screenshot on Windows 11 in four steps, then covers what to do when the recognition comes out wrong.
Step 1 — Take the screenshot at full resolution
Start with a clean capture. Press Win + Shift + S for the Snipping Tool, or use any screenshot utility you already have. Two rules matter for OCR: capture at 100% display scaling or higher, and avoid resizing the image afterwards. Upscaling a blurry screenshot does not add information — the OCR engine still sees mush. If the text is small, zoom the app or page in first, then capture.
Step 2 — Send the image through an OCR engine
Windows 11 itself ships an OCR engine inside the Photos app and PowerToys, but the fastest route for a screenshot is a capture tool with recognition built in. In Snow Shot, take the capture, then choose the text-recognition action in the annotation toolbar. The engine runs on the cropped region and puts the result straight on the clipboard — no file, no upload, no second window.
Step 3 — Paste and check the formatting
Paste into a plain-text editor first (Notepad, or a scratch buffer in your code editor). Word and Google Docs will try to apply smart quotes and autoformatting, which hides OCR mistakes. Check three things: line breaks inside paragraphs, the characters l, 1 and I, and whether minus signs survived as hyphens.
Step 4 — Fix the common failure cases
If words come out garbled, the cause is usually one of four things. Low contrast (grey text on a grey background) — increase contrast before capturing. Anti-aliased small text — zoom in and recapture. Decorative fonts — OCR models are trained mostly on UI and document typefaces. Mixed languages — enable both languages in settings, otherwise the engine silently drops characters it does not expect.
What Windows 11 can and cannot do natively
PowerToys Text Extractor (Win + Shift + T) works well for a quick grab, and the Photos app can copy text from an opened image. Neither keeps a capture history, neither handles scrolling pages, and neither gives you annotation in the same pass. If OCR is a daily task rather than a once-a-month rescue, a dedicated capture tool saves the round trip through Explorer.
OCR for code, tables and receipts
Code needs monospaced text and preserved indentation — turn off smart punctuation so quotes and dashes stay literal. Tables usually lose their grid; expect to paste into a spreadsheet and split columns manually. Receipts and invoices are the hardest common case because of thermal paper contrast and curved surfaces — flatten the paper and shoot straight down if you are capturing with a phone.
Privacy: where does the recognition happen
Local OCR means the image never leaves the machine. Cloud OCR means it does. For screenshots that contain passwords, customer data or internal dashboards, local is the only defensible option. Snow Shot's OCR runs locally; if you switch to a cloud translation feature, only the recognised text is sent, not the screenshot.
Frequently asked
Can I copy text from a screenshot without installing anything? Yes — PowerToys Text Extractor, if you already have PowerToys. Is OCR accurate enough for contracts? For clean, high-resolution scans it is close; always proofread numbers and names. Does OCR work on dark mode screenshots? Yes, but light-on-dark text is recognised slightly worse than dark-on-light, so invert if results look poor.
Copying text from a screenshot without a mouse
If you work keyboard-first: capture with Win + Shift + S style hotkeys, then trigger recognition with a single bound key instead of clicking a toolbar. In Snow Shot you can bind the recognition action to a hotkey, so the whole loop — capture, recognise, paste — runs without leaving the keyboard. This matters more than it sounds: a five-second mouse detour per screenshot adds up to twenty minutes a week for anyone capturing a dozen images a day.
Turning recognised text into something usable
Raw OCR output is a wall of lines. Two quick cleanups make it usable: collapse hard line breaks inside paragraphs (paste into an editor with a join-lines command), and strip the leading spaces that come from indented UI text. If you are feeding the text into an LLM or a search box, remove the line breaks entirely — most tools handle long single-line input better than broken fragments.
Accuracy expectations by content type
Clean UI text: near-perfect. Printed documents scanned at 300 dpi: high, with occasional digit errors. Handwriting: poor, do not rely on it. Screenshots of terminals with dark backgrounds: good if the font is monospaced and at least 12 px. Photos of screens taken with a phone: inconsistent — moiré and glare are the usual culprits.
Want the condensed version? Read the full screenshot OCR guide.