Convert an image to text
Pull the words out of a scan, a screenshot or a photo of a page.
- Private by defaultDeleted within an hour
- Super FastProcess in seconds
- No Sign-upStart instantly
How it works
Add your images
Drag and drop, paste, or pick files. They upload over an encrypted connection.
Pick a quality or a target size
Use the quality slider, or tell it exactly how many KB or MB you need.
Download
Save each file, or grab them all at once as a ZIP.
Add a picture that contains writing and the text is read out of it and shown in a box you can edit, copy or download as a plain text file. It is meant for the everyday cases: a scanned letter, a page photographed from a book, a screenshot of an error message, a receipt, a slide. Several images can be read in one go, and the results downloaded together.
Recognition is done by Tesseract, an open-source engine, with language data for English and Slovenian. Choose the languages that actually appear in the image; picking the right one matters most for letters with diacritics such as č, š and ž. Before reading, small images are enlarged and converted to greyscale, which noticeably improves accuracy on screenshots and low-resolution scans.
Clear, straight, evenly lit printed text is read well. Handwriting, decorative fonts, text on a busy background, and photos taken at an angle or in poor light often are not, and columns or tables may come out in an unexpected order. The engine reports how confident it is, and a warning appears when that is low, so always check the result against the picture. Images are uploaded over an encrypted connection and read on our own servers; the text is not stored.
Resolution decides almost everything
Recognition engines read the shapes of letters, and a shape needs pixels. The long-standing advice for scanning is 300 dots per inch, which makes ordinary body text about thirty pixels tall. At 150 dpi results are still usable for clean print. Below that, letters start to merge: rn is read as m, cl as d. When you photograph a page instead of scanning it, the same rule applies in another form: fill the frame with the text, hold the phone parallel to the paper, and take the picture in good light so the camera does not blur it.
Small screenshots are enlarged before reading, which helps, but a larger and sharper source is always better than a small one made bigger.
What comes back, and what to proofread
The result is plain text: the words and line breaks, without fonts, bold type or layout. Tables arrive as lines of values, and text in two columns may be read straight across. Words that were hyphenated at the end of a line stay split. For a letter or an article, a minute of tidying in the box puts this right.
Numbers deserve real attention. Zero and the letter O, one and a lower-case L, five and S, eight and B look alike to an engine that has no idea what the text means, and a spelling checker will not flag a wrong digit. Compare every amount, date, account number and reference against the picture before you rely on it.
Everyday uses
Most people come here to avoid retyping something. A paragraph from a printed book to quote in an essay. A recipe from a magazine clipping. An error message from a screenshot, to paste into a search engine. The address and telephone number from a business card or a letterhead. The wording of a slide photographed during a talk. A scanned contract that needs to be searchable. Documents like these often contain personal details, which is why the image is deleted after reading and the text is returned to you without being stored.
Frequently asked questions
Which languages are supported?
English and Slovenian. You can select both when an image mixes them.
Can it read handwriting?
Not reliably. The engine is built for printed text; neat block capitals sometimes work, cursive rarely does.
How can I get a better result?
Use a sharp, straight, well-lit image with dark text on a light background, and crop away everything that is not text.
Is the extracted text stored?
No. It is returned to your browser and not kept on our servers.
Can it read German, Italian or Croatian?
Only English and Slovenian language data is installed. Other languages in the Latin alphabet are partly readable, but their special characters come out wrong, so expect to correct them by hand.
Can I get a searchable PDF instead of plain text?
No. The output is text only, to copy or to download as a .txt file.