Image to Text
Turn a photo, scan, screenshot, PDF, or TIFF into text on this device. The file is not uploaded.
Image to Text reads printed text from photos, screenshots, PDFs, and TIFFs on your device. The file stays in your browser — it is not uploaded, stored, or sent to a language model. Copy or download the text when you are done. Made by FreeToolHub.
Image to Text
Drop an image, PDF, or TIFF here, or paste an image
Recognition runs in a worker on this device. The file is not uploaded, and no account is required.
Text
The reader is unsure. Use a sharper picture, leave contrast on, and pick the language the text is written in.
How it works
Choose a picture, a PDF, or a TIFF, paste an image, or take a photo. The tool scales a picture, adds a quiet margin, and can stretch the contrast so faint type is easier to separate from the background. A PDF page that already contains text is read from that text. A scanned PDF or TIFF is drawn in the browser, then Tesseract's LSTM engine reads the pixels in a worker. The language model is one of the twelve shipped with this site, loaded from this site, and kept in the browser for the next page. Printed text and screenshots work well. Handwriting and decorative logos still do not.
Frequently Asked Questions
Does the file leave my device?
No. The file is decoded in this tab and read on this device. Language models are files on this site, not a cloud OCR service. Nothing is uploaded, stored, or sent to a language model.
A chatbot can read images too. Why use this?
A chatbot often reads messy photos better, because a large model is guessing from the whole picture. That guess happens on someone else's server, so the image leaves your device. This tool is the private option: the pixels stay here. It is the right choice for letters, IDs, medical notes, and anything you would not paste into a chat.
What does it read well?
Printed text, scans, and screenshots with clear type, including scanned PDFs and TIFFs. The engine is Tesseract's LSTM model, which is far stronger than the old pattern-matching OCR. It is still weak on handwriting, heavy perspective, and stylized logos. Fill the frame with the text and keep contrast on.
Which languages can it read?
English, Danish, German, French, Spanish, Portuguese, Dutch, Italian, Polish, Turkish, Japanese, and Korean. Pick the language of the text, not the language of this page. Each language is its own model, downloaded once from this site.
Why does the first image take a moment?
The recognition engine and the chosen language model are a few megabytes. They download from this site the first time you read an image, then stay in the browser. Later images start straight away. Switching language loads that language's model once.
Can I photograph a document with my phone?
Yes. Use Take photo, or choose a picture from your camera roll. Hold the page flat, fill the frame, and avoid glare. If the photo is sideways, rotate it before reading. HEIC files from some iPhones need the HEIC to JPG converter first, because browsers cannot decode HEIC.
Which file formats can I open?
PNG, JPEG, GIF, WebP, BMP, AVIF, and SVG, plus PDF and multi-page TIFF. A PDF page that already contains text is read from that text. A scanned page is drawn in the browser and read with OCR. The first 20 pages are read. HEIC still needs the HEIC to JPG converter first.