1 paper · 1 filter
Hong-Phuc Lai, Phong Nguyen, Anh Tran
Pre-trained diffusion models excel at generating high-quality images but remain inherently limited by their native training resolution. Recent training-free approaches have attempt…