How to read a QR code from a screenshot

Use the sharpest version you have: the original screenshot or downloaded image is preferable to a photo of the screen or a repeatedly compressed copy. Keep the square pattern and its surrounding blank border. Cropping tightly through that border, drawing an arrow across the code, or shrinking its modules can make an otherwise valid code difficult to read.

If a web page displays the code at a small size, obtain a larger original through a legitimate download or recapture it at a useful rendered size. Enlarging a blurred bitmap cannot recover missing square boundaries. A screenshot captures what was visible at that moment, so a time-sensitive code may already be expired by the time somebody uses its decoded contents.

QR decoding is not QR generation. It extracts a payload that already exists in the image; it does not create a replacement code, recover an obscured token, or prove the identity of the party that supplied it.

1. Capture the code with the surrounding border, or open the original image in the editor. Avoid adding a highlight, blur, or decoration over the pattern.
2. Choose Image actions → Review text and QR…. Wait for the recognition job to finish instead of repeatedly starting another request.
3. Look for the QR codes result in the Text and QR recognition window. QR payloads are shown separately from ordinary OCR text; a visible URL recognized as printed text is not necessarily the code's decoded content.
4. Use Copy QR contents when the displayed result is the one you intended. Paste it into a plain text field if you need to inspect a long payload completely.
5. Decide the next step yourself. For a URL, check the destination and context before navigating. For other payloads, use the appropriate trusted application only after understanding what the copied text contains.

Suppose a colleague sends a screenshot of a fictional equipment setup sheet. The image includes a QR code beside a printed device name. Open the screenshot, decode it, and compare the payload with the expected setup domain before doing anything else.

A code might contain a URL with a long query string, while the printed text nearby contains a shorter marketing address. Keep the two results distinct. If the decoded host is unexpected, ask the sender through an existing trusted channel rather than treating the nearby logo as proof.

If there are two codes, inspect the image and results carefully before sending either one onward. Klips orders detected QR results by their image position, but a text list does not replace visually identifying which code belongs to which instruction. This example illustrates a review method; it is not a claim that a live device or account was tested.

A successful decode yields a readable payload in the QR section. Confirm it survives copying exactly, including punctuation and any structured fields you actually need. A result copied into the wrong chat or document is still a disclosure, even when recognition itself ran locally.

Some codes carry temporary login links, invitation tokens, contact details, or network configuration. Review the full payload and the original image's sensitivity before sharing either. Copying does not remove the code from the source screenshot, and closing the recognition window does not erase every clipboard or synchronized-history copy.

A decoded URL can also redirect after navigation. The review step is useful, but it does not make the remote site safe or guarantee that its current content matches the screenshot's description.

Check whether the image actually contains a QR code rather than another barcode format or a decorative square. Klips's QR route is not a promise to decode every barcode family. Try an undamaged original and remove only irrelevant surrounding content, leaving the code's border intact.

For a tall capture, keep native source resolution rather than first reducing the whole page to a tiny overview. If the code remains unreadable, request the original link through a trusted route. Do not reconstruct missing modules, submit private screenshots to an arbitrary decoding website, or interpret the absence of a result as evidence that the code is harmless.