AdaState：用於串流影片生成的自我演進錨點

摘要

自回归视频扩散模型通过顺序生成帧来产生流式视频，每一帧块都基于先前生成的内容进行条件生成。这些模型在结构上锚定于第一帧：其键-值表示在注意力缓存中占据特权位置，并在整个生成过程中充当主要场景参考。作为缓存中最干净、出错最少的位置，这一锚点吸引了不成比例的注意力，从而抑制了视频的动态性，并将场景构图锁定在初始视角上，即使场景自然演变也是如此。其结果是生成时间上浅层的视频，其中运动、镜头移动和场景推进被削弱，而静态一致性得到强化。为解决这一问题，我们用自适应状态取代静态锚点——这是一种隐藏潜变量，模型在每一帧块中与内容一同进行去噪处理，但从不渲染。模型不再参考冻结的第一帧，而是通过同时关注先前状态与当前内容，在每一步生成自身的场景锚点，从而产生一个随生成内容演变的参考。与编码绝对时间概念的标准视频生成不同，我们的表述将时间视为相对的：无论生成进展到哪一步，每一步生成都看到相同的位置结构，且每一帧块的状态转移完全相同。这些特性共同在生成过程中引入了一种递归机制，其中去噪充当转移函数，KV缓存充当载体，无需任何外部模块。实验表明，自适应状态显著改善了视频动态性，使得生成视频中能够呈现更丰富的运动和自然的场景推进。

English

Autoregressive video diffusion models generate streaming video by producing frames sequentially, conditioning each chunk on previously generated content. These models are structurally anchored to the first frame: its key-value representation occupies a privileged position in the attention cache and serves as the primary scene reference throughout generation. As the cleanest and most error-free position in the cache, this anchor draws disproportionate attention, suppressing video dynamics, and locking scene composition to the initial viewpoint even as the scene naturally evolves. The result is a temporally shallow video in which motion, camera movement, and scene progression are dampened in favor of static consistency. To address this, we replace the static anchor with an adaptive state, a hidden latent that the model denoises alongside content at every chunk but never renders. Rather than referencing a frozen first frame, the model generates its own scene anchor at each step by attending to both the previous state and the current content, producing a reference that evolves with the generated content. Unlike standard video generation, which encodes an absolute notion of time, our formulation treats time as relative: every generation step sees the same positional structure regardless of how far generation has progressed, and the state transition is identical at every chunk. Together, these properties introduce a recurrence into the generation process, where denoising serves as the transition function, and the KV cache serves as the carrier, requiring no external module. Experiments demonstrate that the adaptive state substantially improves video dynamics, enabling richer motion and natural scene progression within generated videos.