Skip to content

Source paper

HiAR: Efficient Autoregressive Long Video Generation via Hierarchical Denoising

Kai Zou, Dian Zheng, Hongbo Liu and 3 more

Kai Zou, Dian Zheng, Hongbo Liu, Tiankai Hang, Bin Liu, Nenghai Yu

arXiv 2603.08703 · 202624 upvotes on Hugging Face

Research ideas from this paper (3)

Generated by AI. The paper’s authors did not write these.

AI summary

This paper proposes HiAR, an autoregressive framework for long video generation that mitigates distribution drift by utilizing hierarchical denoising. By conditioning generated blocks on context at matched noise levels instead of fully clean frames, it effectively reduces error propagation while enabling pipelined parallel inference.

How it works (AI summary)

The core methodology relies on a hierarchical causal generation schedule where all blocks are generated across denoising steps simultaneously. To prevent motion collapse during training, this is paired with a self-rollout distillation process utilizing a Distribution Matching Distillation (DMD) loss and a bidirectional forward-KL regularizer.