Video-Mirai: Autoregressive Video Diffusion Models Need Foresight

ArXi:2606.03971v1 Announce Type: new Causal video generators must predict from the past, but they need not, however, only asks each causal state to explain the present. This creates what we call a representation-level planning gap: states that fit the current segment may discard identity, layout, and motion information needed for a consistent future. We