Abstract: Large diffusion models demonstrate remarkable zeroshot capabilities in novel view synthesis from a single image. However, these models often face challenges in maintaining consistency across ...