Rolling Forcing：实时自回归长视频扩散
Rolling Forcing: Autoregressive Long Video Diffusion in Real Time
AI总结：
针对流式长视频生成中的误差累积问题，本文提出 Rolling Forcing，通过多帧联合去噪、attention sink 和扩展窗口少步蒸馏，在单 GPU 上实现数分钟实时长视频生成。
摘要：
流式视频生成是交互式世界模型和神经游戏引擎的基础组成部分之一，旨在生成高质量、低延迟且时间一致的长视频流。然而，现有大多数工作都存在严重的误差累积问题，往往会在长时间跨度上显著降低生成流式视频的质量。我们设计了 Rolling Forcing，这是一种新颖的视频生成技术，能够以最小的误差累积实现流式长视频生成。Rolling Forcing 包含三项新颖设计。首先，我们不再迭代采样单个帧——这种做法会加速误差传播——而是设计了一种联合去噪方案，同时对多个具有逐级递增噪声水平的帧进行去噪。该设计放宽了相邻帧之间严格的因果关系，有效抑制了误差增长。其次，我们将 attention sink（注意力汇聚）机制引入长时间跨度流式视频生成任务，使模型能够保留初始帧的键值状态作为全局上下文锚点，从而增强长期全局一致性。第三，我们设计了一种高效训练算法，可在大幅扩展的去噪窗口上实现少步蒸馏。该算法在非重叠窗口上运行，并缓解了以自生成历史为条件时的暴露偏差。大量实验表明，Rolling Forcing 能够在单块 GPU 上实时流式生成数分钟长的视频，同时显著减少误差累积。
英文摘要：
Streaming video generation, as one fundamental component in interactive world models and neural game engines, aims to generate high-quality, low-latency, and temporally coherent long video streams. However, most existing work suffers from severe error accumulation that often significantly degrades the generated stream videos over long horizons. We design Rolling Forcing, a novel video generation technique that enables streaming long videos with minimal error accumulation. Rolling Forcing comes with three novel designs. First, instead of iteratively sampling individual frames, which accelerates error propagation, we design a joint denoising scheme that simultaneously denoises multiple frames with progressively increasing noise levels. This design relaxes the strict causality across adjacent frames, effectively suppressing error growth. 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 and thereby enhances long-term global consistency. Third, we design an efficient training algorithm that enables few-step distillation over largely extended denoising windows. This algorithm operates on non-overlapping windows and mitigates exposure bias conditioned on self-generated histories. Extensive experiments show that Rolling Forcing enables real-time streaming generation of multi-minute videos on a single GPU, with substantially reduced error accumulation.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
