{\Phi}-Noise: Training-Free Temporal Video Conditioning via Phase-Based Noise Manipulation 事件

{\Phi}-Noise: Training-Free Temporal Video Conditioning via Phase-Based Noise Manipulation arXiv:2605.24509v1 Announce Type: new Abstract: Latent video diffusion models generate videos by progressively transforming Gaussian noise into realistic samples conditioned on text or visual inputs. However, existing conditioning methods often require additional training and computational overhead. Motivated by recent findings on the importance of frequency components in generative models, we propose a s