Training-free Diffusion Model Adaptation for Variable-Sized Text-to-Image Synthesis

Diffusion models (DMs) have recently gained attention with state-of-the-art performance in text-to-image synthesis.