Abstract: Advanced diffusion models have made notable progress in text-to-image compositional generation. However, it is still a challenge for existing models to achieve text-image alignment when ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results
Feedback