When you need OCR
Sometimes the text you need is stuck inside an image: a photo of a book page, notes on a whiteboard, a receipt, or a screenshot. Retyping takes time and invites mistakes. OCR (optical character recognition) reads the image and gives you text you can copy and edit.
Prepare the photo first
The result depends heavily on the quality of the image. A few small things help:
- Shoot straight on, not at an angle.
- Use even light with no shadow across the writing.
- Keep the text large and sharp. Move the camera closer instead of using digital zoom.
- Crop out what you don't need so only the text area is left.
How to use Velfile's OCR tool
- Open Image to Text (OCR) and choose your image (JPG, PNG, or WebP, up to 20 MB).
- Pick the language that matches the text in the image.
- Press Extract text, then wait until it finishes.
- Press Copy text or Download .txt to save it as a plain text file.
It all runs in your browser, so your image is not sent to Velfile's servers. The first time, your browser downloads the OCR engine and language data, so you need an internet connection and it takes a little longer.
Pick the right language
An OCR engine recognizes letters based on a language. If the text is in Indonesian but you pick English, words are often misread. Indonesian, English, and Spanish are available right now, one language per run.
What to expect
OCR works best on clear printed text. You still need to check the result:
- Handwriting often doesn't read well.
- Numbers and names can be mixed up, like the letter O and the digit 0. Double-check the important ones.
- Layout such as columns and tables is not kept, only the text.
- Blurry or tilted photos lower the quality. Retaking the photo is faster than fixing the result.
Cleaning up the result
OCR output is rarely perfect straight away. Paste it into Word or Google Docs, then:
- Join broken lines. Text often breaks where the image breaks a line, not at the end of a sentence.
- Check punctuation. Small dots and commas sometimes go missing or appear where they shouldn't.
- Check the numbers. Compare amounts, dates, and reference numbers with the original image.
Common problems
- No text was read. The image may be too blurry or the language may not match. Retake the photo or change the language.
- It takes a long time. A very large image or one full of text needs time. Crop it first with Crop Image.
- A message says the engine could not load. Usually the connection dropped while downloading the OCR engine. Reload the page and try again.
If the source is a scanned PDF
A scanned PDF is just a set of images. This OCR tool doesn't read PDFs directly yet, but you can turn the pages into images with PDF to JPG and then run the images through OCR one by one. If the PDF turns out to have real text, just use PDF to Text.
What to do next
Once the text is cleaned up, you can turn it back into a document with Word to PDF.
Last updated: 25 October 2026