HiAR: Efficient Autoregressive Long Video Generation via Hierarchical Denoising

• URL: http://arxiv.org/abs/2603.08703v1
• Date: Mon, 09 Mar 2026 17:58:16 GMT
• Title: HiAR: Efficient Autoregressive Long Video Generation via Hierarchical Denoising
• Abstract summary: HiAR is a hierarchical denoising framework that reverses the conventional generation order.<n>It performs causal generation across all blocks at every denoising step, so that each block is always conditioned on context at the same noise level.<n>On VBench (20s generation), HiAR achieves the best overall score and the lowest temporal drift among all compared methods.
• Score: 52.237486207964245
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Autoregressive (AR) diffusion offers a promising framework for generating videos of theoretically infinite length. However, a major challenge is maintaining temporal continuity while preventing the progressive quality degradation caused by error accumulation. To ensure continuity, existing methods typically condition on highly denoised contexts; yet, this practice propagates prediction errors with high certainty, thereby exacerbating degradation. In this paper, we argue that a highly clean context is unnecessary. Drawing inspiration from bidirectional diffusion models, which denoise frames at a shared noise level while maintaining coherence, we propose that conditioning on context at the same noise level as the current block provides sufficient signal for temporal consistency while effectively mitigating error propagation. Building on this insight, we propose HiAR, a hierarchical denoising framework that reverses the conventional generation order: instead of completing each block sequentially, it performs causal generation across all blocks at every denoising step, so that each block is always conditioned on context at the same noise level. This hierarchy naturally admits pipelined parallel inference, yielding a 1.8 wall-clock speedup in our 4-step setting. We further observe that self-rollout distillation under this paradigm amplifies a low-motion shortcut inherent to the mode-seeking reverse-KL objective. To counteract this, we introduce a forward-KL regulariser in bidirectional-attention mode, which preserves motion diversity for causal inference without interfering with the distillation loss. On VBench (20s generation), HiAR achieves the best overall score and the lowest temporal drift among all compared methods.

Related papers

• In-Context Forcing: Uncovering Context Effects in Autoregressive Video Diffusion [52.18702395026718]
  We introduce In-Context Forcing, a progressive autoregressive paradigm that utilizes contexts with decreasing noise levels.<n>By applying less masking to distant frames and more masking to adjacent ones, this approach provides adaptive guidance.<n>Our method significantly outperforms state-of-the-art approaches in both visual fidelity and inference speed.
  arXiv  Detail & Related papers  (2026-08-05T13:54:31Z)
• LeapTalk: Breaking the Latency-Quality Trade-off in Talking Head Generation [28.818825505825476]
  We propose LeapTalk, a novel framework that achieves stable and real-time talking-head generation with a single forward step.<n>Our method achieves high-fidelity and temporally consistent video generation with only 1 step at up to 200 FPS.
  arXiv  Detail & Related papers  (2026-07-29T14:44:57Z)
• Cycle-World: Mitigating Error Accumulation in Long-term Video World Models via Reverse-Prediction Cycle Consistency [123.12549538776109]
  Cycle-World is a framework designed for stable and temporally consistent long-video generation.<n>Our approach tackles error drift by enforcing strict temporal reversibility across both the training and inference phases.<n>Experiments on the VBench benchmark demonstrate that Cycle-World's dual-phase synergy significantly mitigates error drift.
  arXiv  Detail & Related papers  (2026-07-13T17:27:13Z)
• BlockVLA: Accelerating Autoregressive VLA via Block Diffusion Finetuning [41.5997751218601]
  BlockVLA is a framework that adapts pretrained AR backbones into an efficient discrete diffusion policy.<n>We conduct extensive evaluations on the LIBERO and SimplerEnv benchmarks.<n>Our model exhibits superior training efficiency, with success rates converging substantially faster than baselines.
  arXiv  Detail & Related papers  (2026-05-13T11:37:51Z)
• OmniForcing: Unleashing Real-time Joint Audio-Visual Generation [51.031726911410594]
  Recent joint audio-visual diffusion models achieve remarkable generation quality but suffer from high latency.<n>We propose OmniForcing, the first framework to distill an offline, dual-stream bidirectional diffusion model into a high-fidelity streaming autore generator.
  arXiv  Detail & Related papers  (2026-03-12T08:17:36Z)
• Streaming Autoregressive Video Generation via Diagonal Distillation [50.13573884115673]
  Autoregressive models offer a natural framework for sequential frame synthesis but require heavy computation to achieve high fidelity.<n>We propose Diagonal Distillation to exploit temporal information across both video chunks and denoising steps.<n>Our method generates a 5-second video in 2.61 seconds (up to 31 FPS), achieving a 277.3x speedup over the undistilled model.
  arXiv  Detail & Related papers  (2026-03-10T10:45:24Z)
• Rejection Mixing: Fast Semantic Propagation of Mask Tokens for Efficient DLLM Inference [58.189320101488725]
  DLLMs promise fast non-autoregressive inference but suffer a severe quality-speed trade-off in parallel decoding.<n>We address this by integrating continuous representations into the discrete decoding process, as they preserve rich inter-position dependency.<n>We propose ReMix, a framework that introduces a novel Continuous Mixing State as an intermediate between the initial masked state and the final decoded token state.
  arXiv  Detail & Related papers  (2026-02-26T11:08:11Z)
• Causality in Video Diffusers is Separable from Denoising [38.11368818072302]
  Causality underlies many complex generative processes, including videos, language, and robot trajectories.<n>Current causal diffusion models entangle temporal reasoning with iterative denoising, applying causal attention across all layers, at every denoising step, and over the entire context.<n>We show that the causal reasoning in these models is separable from the multi-step denoising process.<n>Motivated by these findings, we introduce Separable Causal Diffusion (SCD), a new architecture that explicitly decouples once-per-frame temporal reasoning, via a causal transformer encoder, from multi-step frame-wise rendering
  arXiv  Detail & Related papers  (2026-02-10T18:57:21Z)
• LoL: Longer than Longer, Scaling Video Generation to Hour [50.945885467651216]
  This work achieves the first demonstration of real-time, streaming, and infinite-length video generation with little quality decay.<n>As an illustration, we generate continuous videos up to 12 hours in length, which, to our knowledge, is among the longest publicly demonstrated results in streaming video generation.
  arXiv  Detail & Related papers  (2026-01-23T17:21:35Z)
• Adaptive Begin-of-Video Tokens for Autoregressive Video Diffusion Models [11.913945404405865]
  Most video diffusion models (VDMs) generate videos in an autoregressive manner, generating subsequent iteration frames conditioned on previous ones.<n>We propose Adaptive Begin-of-Video Tokens (ada-BOV) for autoregressive VDMs.
  arXiv  Detail & Related papers  (2025-11-15T08:29:14Z)
• CANDI: Hybrid Discrete-Continuous Diffusion Models [36.61898210733147]
  We show how noise corrupts discrete data through two mechanisms: discrete identity corruption and continuous rank degradation.<n>We propose CANDI, a hybrid framework that decouples discrete and continuous corruption.<n>This unlocks the benefits of continuous diffusion for discrete spaces.
  arXiv  Detail & Related papers  (2025-10-26T03:24:31Z)
• Rolling Forcing: Autoregressive Long Video Diffusion in Real Time [86.40480237741609]
  Rolling Forcing is a novel video generation technique that enables streaming long videos with minimal error accumulation.<n>Rolling Forcing comes with three novel designs. First, instead of iteratively sampling individual frames, which accelerates error propagation, we design a joint denoising scheme.<n>Second, we introduce the attention sink mechanism into the long-horizon stream video generation task, which allows the model to keep key value states of initial frames as a global context anchor.<n>Third, we design an efficient training algorithm that enables few-step distillation over largely extended denoising windows.
  arXiv  Detail & Related papers  (2025-09-29T17:57:14Z)
• SparseD: Sparse Attention for Diffusion Language Models [98.05780626106555]
  diffusion language models (DLMs) offer a promising alternative to autoregressive models (ARs)<n>Existing open-source DLMs suffer from high inference latency.<n>We propose SparseD, a novel sparse attention method for DLMs.
  arXiv  Detail & Related papers  (2025-09-28T18:10:10Z)
• Flow Matching in the Low-Noise Regime: Pathologies and a Contrastive Remedy [19.64681328133978]
  Flow matching provides a continuous-time formulation for generative modeling and representation learning.<n>We show that this framework suffers from a fundamental instability in the low-noise regime.<n>We propose Local Contrastive Flow (LCF), a hybrid training protocol that replaces direct velocity regression with contrastive feature alignment at small noise levels.
  arXiv  Detail & Related papers  (2025-09-25T09:36:41Z)
• DiffusionAD: Norm-guided One-step Denoising Diffusion for Anomaly Detection [80.20339155618612]
  DiffusionAD is a novel anomaly detection pipeline comprising a reconstruction sub-network and a segmentation sub-network.<n>A rapid one-step denoising paradigm achieves hundreds of times acceleration while preserving comparable reconstruction quality.<n>Considering the diversity in the manifestation of anomalies, we propose a norm-guided paradigm to integrate the benefits of multiple noise scales.
  arXiv  Detail & Related papers  (2023-03-15T16:14:06Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.