通过智能选择计算帧，实现视频生成加速。
Not All Frames Deserve Full Computation: Accelerating Autoregressive Video Generation via Selective Computation and Predictive Extrapolation
- 提出三模式调度：缓存、预测、重算，动态平衡效率与精度。
- 在MAGI-1和SkyReels-V2上实现最高4.73倍加速，质量无损。
- 适合追求高效视频生成且不希望训练的开发者使用。
自回归视频扩散模型虽能生成长视频，但因重复多步去噪而代价高昂。现有无需训练的加速方法依赖二元缓存或重算决策，忽略中间状态；同时异步自回归调度为共生成帧分配不同噪声水平，但现有方法仍统一处理有效区间。为此，我们提出SCOPE，一种无需训练的高效自回归视频扩散框架。SCOPE引入三模式调度（缓存、预测、重算），通过噪声级泰勒外推进行预测，填补重用与重算之间的空白，并基于误差传播分析提供显式稳定性控制。进一步引入选择性计算，仅对活跃帧区间执行运算。在MAGI-1和SkyReels-V2数据集上，SCOPE实现最高4.73倍加速，且生成质量与原模型相当，优于所有无需训练的基线方法。
原文摘要 · Abstract (English)
Autoregressive (AR) video diffusion models enable long-form video generation but remain expensive due to repeated multi-step denoising. Existing training-free acceleration methods rely on binary cache-or-recompute decisions, overlooking intermediate cases where direct reuse is too coarse yet full recomputation is unnecessary. Moreover, asynchronous AR schedules assign different noise levels to co-generated frames, yet existing methods process the entire valid interval uniformly. To address these AR-specific inefficiencies, we present SCOPE, a training-free framework for efficient AR video diffusion. SCOPE introduces a tri-modal scheduler over cache, predict, and recompute, where prediction via noise-level Taylor extrapolation fills the gap between reuse and recomputation with explicit stability controls backed by error propagation analysis. It further introduces selective computation that restricts execution to the active frame interval. On MAGI-1 and SkyReels-V2, SCOPE achieves up to 4.73x speedup while maintaining quality comparable to the original output, outperforming all training-free baselines.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。