AI RESEARCH
AnyFlow: Any-Step Video Diffusion Model with On-Policy Flow Map Distillation
arXiv CS.AI
•
ArXi:2605.13724v1 Announce Type: cross Few-step video generation has been significantly advanced by consistency distillation. However, the performance of consistency-distilled models often degrades as sampling steps are allocated at test time, limiting their effectiveness for any-step video diffusion. This limitation arises because consistency distillation replaces the original probability-flow ODE trajectory with a consistency-sampling trajectory, weakening the desirable test-time scaling behavior of ODE sampling. To address this limitation, we