TY  - RPRT
TI  - CompAlign: Improving Compositional Text-to-Image Generation with a Complex Benchmark and Fine-Grained Feedback
AU  - Yixin Wan
AU  - Kai-Wei Chang
PY  - 2025
UR  - https://arxiv.org/abs/2505.11178
ID  - 2505.11178
ER  -