AI RESEARCH
Accelerating Video Inverse Problem Solvers with Autoregressive Diffusion Models
arXiv CS.LG
•
ArXi:2605.20624v1 Announce Type: cross Diffusion models provide powerful priors for zero-shot video inverse problems, but their real-time deployment is hindered by two inefficiencies: high initial latency caused by holistic video restoration, and low throughput resulting from multiple VAE passes to enforce measurement consistency in pixel space. To overcome these limitations, we propose Autoregressive Video Inverse problem Solver (AVIS). The AVIS framework leverages autoregressive video diffusion models to re videos in a streaming manner, naturally eliminating latency bottlenecks.