1 paper
Kirill Koltsov, Aleksandr Gushchin, Anastasia Antsiferova +1
Recent text-to-image models have improved global realism, but text rendering remains a persistent failure mode: images may look convincing overall, yet local typography often conta…