No watermark, no account, no limits
Free foreverNo sign upNo watermark
Pull the words out of a screenshot, a photo or a scan, and copy them as real editable text.
Drop your image here
or click to choose a file
JPG, PNG, WebP, GIF or BMP · photo, screenshot or scan
Starting the recognition engine…
Text found
Runs entirely on your device — your image is never uploaded. Safe for invoices, contracts and ID documents.
Nothing uploadedWorks offline after first runNo sign up
No account. No trial. No watermark. No "upgrade to download."
Three things worth knowing before you use it.
The whole thing runs on your device. Your image never reaches a server, which is why it is safe to use on an invoice, a contract or an ID document.
Crisp on-screen text is the easiest thing to read accurately. A clear screenshot is near-perfect; a phone photo of a curved page at an angle is much harder.
What comes back is text you can copy, paste, search and edit — not a picture of text. Download it as a .txt file if you want it saved.
No account, no install, nothing uploaded.
Drag it in or pick it from your phone or computer.
One click. The first run downloads the engine, so give it a few seconds.
Copy to the clipboard, or download it as a text file.
The everyday reasons.
One file or two hundred. No daily counter, no paid tier hiding behind it.
Fair question. Here's the honest version.
Upload a JPG, PNG, WebP, GIF or BMP, pick the language if it is not English, and click Extract text. The words come back as editable text you can copy or download. There is no account, no watermark and no limit.
Drop the screenshot straight in — screenshots are the best case for this. On-screen text is crisp and evenly lit, so recognition is usually close to perfect, unlike a photograph where focus, angle and shadow all get in the way.
Yes, and scans work well when they are straight and reasonably sharp. A flatbed scan at 300 DPI reads far better than a phone photo of the same page. If the scan is skewed, straighten it before uploading — even a few degrees of rotation costs accuracy.
This page works in Safari with nothing to install — tap the upload area and pick the photo. iOS also has Live Text built in: press and hold on text in a photo and it becomes selectable. That is quicker for a phrase; this is better for a whole page, and it gives you a file you can save.
No. The recognition runs entirely in your browser, so the image never leaves your device. That is the reason this is safe for an invoice, a contract, a payslip or an ID document — there is nothing on our end to store or expose.
The recognition engine downloads to your browser the first time you use it, which takes a few seconds. After that it is cached, so later extractions start immediately. This is the trade-off for doing the work locally instead of on a server.
Blur, low contrast, unusual or decorative fonts, handwriting, text at an angle, and heavy shadow. Printed text photographed straight-on in good light reads almost perfectly. Handwriting is unreliable for anything beyond neat block capitals.
Yes — pick the language from the dropdown before extracting. Choosing the right one matters more than people expect, because the engine uses the language to resolve ambiguous characters. Extracting Spanish with English selected will produce noticeably worse results.
Same rules on every one: free, no sign up, no watermark.
More reading: how OCR works and Tesseract, the engine behind it.