Are the columns of a receipt respected?
Yes: with the “Receipt, invoice, table” type, each printed line stays a line, and the columns are separated by tabs.
Which languages are recognised?
French, Dutch, English, German, Spanish, Italian, Portuguese and Polish.
What about a scanned PDF?
Use the “OCR: make a PDF searchable” tool of ElniDocs (elnidocs.com), the twin site for documents: it keeps the PDF's appearance.
How do I get better text recognition?
Photograph the document flat, well lit and sharp, framing only the text. A small image is enlarged up to 2 times before reading; still check characters that are easy to confuse, such as 0 and O or 1 and l.
Can I extract text from an iPhone HEIC photo?
Convert it first with “HEIC to JPG”: text recognition accepts JPG, PNG and WebP images.
What needs to be downloaded to read the text in an image?
The first time, the recognition engine (about 4 MB) and the model for each language chosen, from 0.7 MB for French to 3 MB for Dutch. They come from ElniFrame and then stay cached in the browser.
Are my files sent to a server?
No. Everything is processed in your browser, on your device: no file is sent anywhere, neither to ElniFrame nor to anyone else. You can check this by going offline once the page has loaded: the tool keeps working (for an AI or text recognition tool, once its model has been downloaded, the first time).
Is it free? Do I need an account?
It is free, with no sign-up, no adverts and no watermark added to your files. The only limit is your device's memory (up to 500 MB per image and 4 GB per video).