AI RESEARCH

DySink: Dynamic Frame Sinks for Autoregressive Long Video Generation

arXiv CS.CV

ArXi:2605.21028v1 Announce Type: new Autoregressive long video generation often adopts bounded-memory streaming for efficiency, typically combining local windows for short-term continuity with static early-frame sinks as long-range anchors. However, this fixed allocation keeps early frames cached even when the current visual state has substantially diverged from them, while discarding potentially relevant intermediate history.