To copy printed text from a photo, open Image to Text / OCR, choose your image, select its language and click Convert Now. Download the TXT file, then compare it with the original before using the text. This works for a screenshot, a photographed letter or a short scanned PDF; the result is editable plain text.

A useful OCR workflow starts with the source you already have. If a paragraph looks blurred or tiny when you zoom in, recognition may struggle with it too. Getting a clearer image and checking a few important words usually matters more than repeatedly processing the same poor photograph.

Choose a supported source

The tool accepts one PNG, JPG, JPEG or WebP image, or an unencrypted PDF. The upload limit is 20 MiB. Image dimensions must stay within 16 million pixels and 12,000 pixels on either side. A small file in bytes can still contain too many pixels, so check both file size and dimensions when an upload is rejected.

PDFs can contain up to 10 pages. Each page is rendered at 150 DPI for recognition and must fit the same pixel limits. A large drawing sheet may exceed them even when its PDF file is small. For a longer document, use Split PDF to make shorter sections, then process each section separately.

Choose a source containing printed English, Arabic or both. Handwriting, decorative lettering and complicated page layouts may produce incomplete text or errors. The page does not offer a handwriting mode or a language beyond those three selections.

Prepare a photo without changing its meaning

Keep the printed page upright, in focus and evenly lit. Photographing it straight on gives the letters a more consistent shape than a steep angle. For a screenshot, save the original image rather than a smaller preview from a messaging app.

Crop away unrelated material while leaving a little clear space around the text. Keep full lines and punctuation inside the crop. If you need to rotate or crop an existing image, Image Editor provides those controls. Export the prepared image, then open it in the OCR tool.

Tesseract's own image-quality guidance explains that skew, noise and unsuitable borders can reduce recognition quality. Straightening text lines and using a reasonable border are practical preparation steps. Zutilium does not expose the advanced recognition settings discussed in that documentation; its visible setting is the text language.

Run OCR and download the text

1. Open Image to Text / OCR and choose the source file in the upload area.
2. Set Text language to English, Arabic or English + Arabic.
3. Click Convert Now and wait for processing to finish.
4. Click Download to save the generated TXT file.
5. Open the text beside the original and correct mistakes.

For an English letter with an Arabic address, choose English + Arabic. For a page containing only Arabic text, choose Arabic. Match the setting to the printed content rather than the language of your browser.

Selecting a file prepares the page for conversion. Clicking Convert Now uploads it to the server for OCR. Recognition is not performed locally in your browser. Processing has a two-minute deadline, and the resulting download is temporary, so save it promptly.

Check a concrete example

Imagine a photographed delivery note containing an order identifier, two product names and a total. After downloading the text, check the identifier and total first: those are the parts where a single incorrect character could change what the document means. Then check each product line against the photograph.

If the source says “Order 1048” and “Total 120.00”, verify every digit and the decimal point. Do not assume that a readable paragraph means all numbers are correct. Names, email addresses and reference codes also deserve a direct comparison with the source.

For a two-page scanned letter, check that the first page's closing line appears before the second page's opening line. PDF output keeps page order and inserts page boundaries in the plain-text file. An empty page can leave an empty section between those boundaries.

Understand the output before reusing it

The download is UTF-8 TXT. It contains recognized text, rather than a photograph of the page. It does not preserve the original fonts, images, table cells or exact layout, and it does not add a searchable text layer to the source PDF.

A table may arrive as text with spacing or an unexpected reading order. Reconstruct its rows and columns only after checking the original. If you want a Word document, you can add Markdown headings or lists to the corrected text and use Markdown to Word to export DOCX. OCR itself returns TXT.

Fix an empty result or a failed conversion

“No readable text was found” can mean that the source is blank, the text is too difficult to distinguish or the selected language is unsuitable. Check the language, use a clearer source and try a focused crop containing complete printed lines.

If the tool reports a page or pixel limit, reducing the PDF's file size alone may not resolve it. Split the document or create a smaller image with sufficient readable detail. If processing takes too long, try one image or fewer PDF pages. Keep the original, the corrected text and your downloaded result together so you can check the transcription again when needed.