1 paper
Yaohou Fan, Qingzhong Wang, Yongsong Huang +3
Current visual text generation models struggle with the trade-off between text accuracy and overall image coherence. We find that achieving high text accuracy can reduce aesthetic…