running but not processing

[DonKam-ctrl]

i've looked through the manual for installation process and 100% sure it is correct but on the debug tab the original and processed frame are the same? has anyone had any luck with all the translation tools on github or just me

[tomkam1702]

1. Use these two settings in the Settings tab.

2. You should see both the original and processed (here black-and-white) images in the Debugging tab. You can save them to your local disk, if you like. The processed image is the source text area pre-processed for OCR. In the Tesseract OCR Results section, you can see how well Tesseract can recognize/extract text from the processed image.

[DonKam-ctrl]

korean to english or other way around doesn't work could be that, tried changing module from marian to gemini but i cant get an api key the website wont generate one

[tomkam1702]

1. MarianMT models have only the Korean-to-English pair. Here's a sample screenshot from a Netflix series translated with the MarianMT model and Gemini 2.0 Flash for OCR. Tesseract can produce poor or no results if the background is challenging:

[DonKam-ctrl]

could be that, trying to use on a social media messaging client and also in a game but its not recognising it very strange

[tomkam1702]

One more example. This is with Gemini 2.0 Flash for translation and Tesseract for OCR. Tesseract definitely works for Korean, just be sure to disable "Remove Trailing Garbage" to correctly process text strings with no punctuation at the end.

Also: When installing Tesseract, check all options, install all features to have support for all supported languages and scripts.

[DonKam-ctrl]

installing all options in tess and disabled the garbage setting starts to work but its printing gibberish text