I’m encountering a persistent issue with DALL·E and image generation tools in ChatGPT where the Korean character “엌” is never rendered correctly. It either produces gibberish strokes, Latin-looking characters, or misassembled components like “ㅇ ㅏ F”.
This is part of a broader issue where the AI fails to handle Hangul syllables composed of complex consonant clusters or rarely used final consonants.
I believe this is a systemic problem related to the lack of training data for complete Hangul syllables and poor support for Korean typography in the image generation model.
Please escalate this to the font handling and engineering teams so that Hangul support can be improved.
This issue significantly affects usability for Korean users and should not be overlooked.