Where Culture Fades: Revealing the Cultural Gap in Text-to-Image Generation

Chuancheng Shi, Shangze Li, Shiming Guo +9

Multilingual text-to-image (T2I) models have advanced rapidly in terms of visual realism and semantic alignment, and are now widely utilized. Yet outputs vary across cultural conte…