What an image to text converter does
Image to text conversion, often called OCR, turns the letters in a picture into real characters. Once the text is extracted you can search it, edit it, paste it into a document or a message, or read it aloud with a screen reader, none of which is possible while the words are only pixels.
Adding translation to the same step saves a round trip. Instead of copying the extracted text into a separate translator, you choose the language before you start and get both versions side by side.
How to use the image to text translator
Upload a JPG, PNG or WebP image up to 10 MB. Leave the language menu on No translation to extract the text only, or choose one of the 19 target languages. Press Extract text and the result appears in a few seconds, with the detected language shown above it.
Each block has a Copy button, so the extracted text and the translation can go straight into an email, a document or a chat. Line breaks are kept, which keeps addresses, lists and menus readable.
Pictures that work well
Printed text gives the best results: menus, signs, product labels, book pages, slides, receipts and screenshots of apps or websites. Clear, evenly lit photos taken straight on are read almost perfectly, even when the text is small, because the image is sent to the model at up to 1,600 pixels on the longest side.
Handwriting works when it is neat and reasonably large, but results vary with the writer. Stylised logos, text curved around objects and words in the far background are the most likely to be missed or misread.
Tips for accurate text extraction
Photograph the text as flat and square as you can, with the whole area in focus and no glare on glossy paper. Crop away everything that is not text, especially busy backgrounds, so the model concentrates on the words you need.
For long documents, one photo per page reads better than a single picture of an open book. If a word comes out wrong, a closer, sharper shot of that area usually fixes it.
Translating the text
The translation keeps the line structure of the original, which makes menus and forms easy to compare line by line. Target languages include English, Spanish, French, German, Portuguese, Italian, Japanese, Korean, simplified and traditional Chinese, Arabic, Hindi and more.
Machine translation is excellent for understanding a text and for everyday use such as travel, shopping and studying. For contracts, medical instructions or official documents, have the translation checked by a qualified person before relying on it.
Everyday uses for an image to text translator
Travellers photograph menus, station signs, tickets and opening hours and read them in their own language. Shoppers check ingredient lists and care labels on imported products. Students turn photos of textbook pages and lecture slides into notes they can search and edit.
At work, the same tool captures text from whiteboards after a meeting, from screenshots of error messages and from scanned letters, and it helps anyone who finds small print hard to read by turning it into text that a screen reader or a larger font can handle.
Copying text out of screenshots
Retyping text from a screenshot is slow and error-prone. Upload the screenshot instead and copy the result: chat messages, social posts, quotes from videos and error messages all come out as editable text in seconds.
For code or commands, compare the result with the original before running it. Spacing, similar-looking characters and special symbols are the most common places for a transcription to differ.
Right-to-left and East Asian scripts
The image to text translator reads many writing systems, including Arabic and Hebrew, which run right to left, and Chinese, Japanese and Korean. Text in those scripts is returned in its natural reading order, ready to be translated into any of the target languages.
Vertical Japanese or Chinese text, such as on a traditional sign, is transcribed column by column. Check the order on complex layouts, and crop to a single column if the sequence looks mixed up.
Tables, forms and complex layouts
Text is returned in reading order, top to bottom and left to right. Simple tables usually come through with one row per line, but complex layouts with merged cells, multiple columns or text running around images can lose some of their structure.
When the layout matters, crop one column or one table at a time. The extracted text is then easier to paste into a spreadsheet or document and rebuild in its original form.
Working with the extracted text
Once the image to text translator has finished, the text behaves like anything you typed yourself. Paste it into a document to edit, into a spreadsheet to sort, into a search box to look something up, or into your notes app to keep. Screen readers and text-to-speech tools can read it aloud.
Results are shown on the page but not stored in your workspace, so copy what you need before you upload the next picture. Running the same image again uses another credit.
Which model reads your image
The image to text translator uses Google Gemini 3.1 Flash-Lite through OpenRouter, with Gemini 2.5 Flash-Lite or OpenAI GPT-5 nano as fallbacks. The model is told to transcribe the text exactly as it appears, without describing the image or correcting the source, and every result passes an automated content-safety check.
Because the image is processed by these providers, avoid uploading highly sensitive documents such as passports, bank statements or medical records. For everyday material like menus, notes and screenshots, the tool is a quick and convenient reader.