Image text translation
A page photographed on a phone, a screenshot of some instructions, a table shot from a report — the text is inside the picture, unselectable and unsearchable. Geyi recognises those characters, translates them, and lays the result out as an editable document.
One thing to be clear about: what you get is a document, not a translated picture. The translation is not painted back onto the original image; it is reflowed into Word, PDF, Markdown or EPUB. That is more useful when you want readable, editable text — but it is not a way to reproduce the image itself.
Supported image formats
PNG, JPEG, WEBP, GIF and TIFF can be uploaded directly. If you have several images, combining them into one PDF first bills on the total page count, which is cheaper than uploading them one at a time — every file carries a 10-credit minimum.
How it is billed
A single image always counts as exactly one page, regardless of file size or how much text it contains. That is 8 credits on high precision and 4 on standard — but because every file bills at least 10 credits, one image on its own costs 10 either way.
So ten images uploaded separately cost 100 credits; the same ten combined into one PDF cost 80 on high precision.
When you only want the text
To pull the characters out without translating, use document extraction: 4 credits a page, a single image still counting as one page, producing Word, PDF, TXT or Markdown in the original language.
Frequently asked questions
Can I get a translated image back?
No. The translation is laid out as an editable document (Word, PDF, Markdown or EPUB) rather than painted back onto the original picture.
What does one image cost?
10 credits. One image counts as one page, and every file bills at least 10 credits, so both precision tiers cost the same for a single image.
How do I keep the cost down?
Combine several images into one PDF before uploading. A ten-page PDF on high precision is 80 credits, against 100 for ten separate uploads.
Can interface text in screenshots be recognised?
Yes — clean screenshots are recognised well. Very small type or heavily compressed images are harder.
Which image formats are supported?
PNG, JPEG, WEBP, GIF and TIFF.
What about comics or vertically set images?
Vertical text is converted to horizontal and the reading order is corrected. But the output is a text document rather than translated text placed back into the artwork, so it suits understanding the content rather than producing a localised edition.
Related
中文版
Geyi is a proprietary AI document platform owned and operated by Nayuta Technology and Culture Company Limited.