Image to Text. Screenshot to Text.

FAQ

How does image to text work here?

Drop a screenshot, photo, or PDF on the box above. Recognition runs in this browser tab with open-source Tesseract OCR compiled to WebAssembly, so the pixels are read on your own device rather than on a remote server. You can also paste from the clipboard or pick a file from disk. When the run finishes, the text lands in an editable box you can fix up, copy in one tap, or save as a plain .txt file. Closing or reloading the tab clears the working copy in memory.

WebP, PNG, and JPEG: what image types work?

All three, plus GIF and BMP images and PDF files. Your browser decodes the file before recognition starts, so the container itself does not change the result. Quality does. A lossless PNG keeps thin interface fonts crisp, while a WebP or JPEG saved at a low setting can smudge small letters and cost you a few characters. Chrome often saves captures as WebP, and those usually read fine. Drop one file or several at once, and you can mix images and a PDF in the same batch. When more than one file returns text, each block is labeled so you can tell them apart.

Can I paste a screenshot instead of saving a file?

Yes. Take the screenshot, then press Ctrl+V or ⌘V anywhere on this page. The image goes straight into the reader, so you can skip saving it to disk and skip naming a temporary file. That path is handy for error dialogs, chat snippets, and quick UI captures. Dropping a file and browsing for one both still work when the picture is already on disk. Paste starts its own run, the same as a drop, so finish or stop the current job before you paste the next one.

Can it read a multi-page PDF?

Yes. Each page is drawn in this tab and then read, one page after another, so a long document does not freeze the browser. The status line names the page being read, and a page that fails is skipped while the rest continue. A Stop button sits next to the progress bar while a run is going. Stopping ends the run at the current page and keeps every line found up to that point, so you can copy what you have instead of starting over.

Do I get a searchable PDF back?

Not yet, and we would rather be straight about it. What you get is the text itself, in an editable box you can copy or save as a .txt file. The tool does not write that text back into the PDF as an invisible layer, so it will not hand you a searchable PDF with the original layout intact. Every page is read as pixels here, even a PDF that already has selectable text in it. If you need the words, this works well. If you need the same document with a text layer added, you need a different kind of tool.

Which languages can it read?

English is the default. You can also pick Spanish, French, German, Portuguese, Italian, Dutch, Chinese (Simplified), Japanese, or Korean before you run the tool. Choose the language first, because the matching recognition data file loads with it. Each language has its own file, so the first read in a language you have not used yet takes a moment longer while that file downloads and caches. Mixed-language screenshots are harder: pick the language that covers most of the text, then fix stray words in the editable box. Switching languages mid-session only fetches the new file once.

Does my screenshot leave my device?

No. The recognition engine is a WebAssembly build that your browser downloads once and then runs inside this tab, so your image or PDF is read where it already sits. There is no upload step for recognition, and we do not store the picture or the extracted text on a server. The file stays in the tab, and closing the tab clears it. The privacy page spells out what the site does and does not collect.

How is this different from an upload converter?

Upload converters send your file to their servers, run recognition there, and send text back. That costs them money for every file, which is why guests often meet a daily quota, a queue at busy times, or a sign-in wall before the good part. Those sites also end up holding a copy of your file for a while, however briefly. Here the engine downloads once and the work happens on your device, so there is no upload to sit through, no quota, and no copy of your screenshot in someone else's storage. A slow connection only slows that first engine download, not the files you read after it. The trade is that heavy jobs lean on your own processor rather than a server farm.

Is it free, and do I need an account?

It is free to use. There is no account, no trial, no paywall, and no daily limit on how many files you read. Ads on the page cover hosting so basic OCR stays open without a paid gate. Reading happens on your device, so extra files cost us nothing and there is nothing for us to meter. Closing the tab does not create a billable event, because there never was one.

Do the ads see my file?

Never. Ads are loaded by Google AdSense in the page chrome, while recognition happens in the tab on your own device. Ad code never receives your image, your PDF, or the text the tool extracted, and nothing about your file is used to choose an ad. That split is intentional. Ad code is not a side channel for your screenshot. If an ad unit covers the drop zone or the copy box on your viewport, say so on the contact page with the URL and device details.

Why is the first read slower than the next one?

The recognition engine is a WebAssembly build, and every language has its own data file. Your browser fetches both before the first read, which is the pause you notice. Those requests carry the engine and the language data only, never your file. The browser then caches them, so the next file in this session starts right away and a later visit skips the download too. Switching to a language you have not used before fetches that one file once.

How accurate is it on handwriting or a low-quality scan?

Printed text is where it does best. Sharp interface screenshots, clean PDF pages, and straight-on photos of typed documents come back cleanest. Handwriting is the weak spot and we would rather say so than oversell it. The engine is trained on printed letterforms, so notes and cursive often come back with gaps or nothing at all. A faded scan, a fax, low contrast, glare, tilt, and stylized display fonts all cost accuracy too. When a result looks rough, crop tight around the text and use the largest copy of the image you have. That helps far more than running the same blurry file again.

Can I edit the text, and should I copy it or download it?

The output box is editable, so you can fix a misread letter or trim stray line breaks before you take the text anywhere. Copy puts the whole box on your clipboard in one tap, which suits a quick paste into a chat, a ticket, or a doc. Download .txt saves the same text as a plain file, which suits long PDF runs and anything you want to keep. Both use exactly what is in the box, edits included, and nothing you type there is sent to us.