Search papers, labs, and topics across Lattice.
LF-MultiDiffusion introduces a novel training-free method for generating spherical panoramas by reformulating latent aggregation as a regularized least-squares problem solved with a Krylov-based iterative solver. This approach allows for denser and more natural mappings between target and reference image spaces, resulting in improved visual quality and consistency with significantly fewer perspective views. The method achieves a remarkable 15.36脳 speedup in inference efficiency compared to existing training-free baselines, making it a compelling advancement in panorama generation techniques.
LF-MultiDiffusion achieves a 15.36脳 speedup in spherical panorama generation while enhancing visual quality and consistency without the need for training.
We propose LF-MultiDiffusion, a training-free panorama generation method that extends MultiDiffusion to support linear projections between target and reference image spaces. Our key idea is to reformulate latent aggregation as a regularized least-squares problem and solve it efficiently with a Krylov-based iterative solver inside the denoising loop. This formulation enables denser and more natural mappings than prior training-free methods, yielding more stable generation with far fewer perspective views. As a result, LF-MultiDiffusion reduces the number of image generator evaluations during denoising and significantly improves inference efficiency. Experiments show that LF-MultiDiffusion achieves better visual quality, text alignment, and panoramic consistency than the strongest training-free baseline, while providing a 15.36$\times$ speedup. Our project page is available at: https://ahykw.github.io/lfmd.
