发表机构
The University of Hong Kong; Tsinghua University（香港大学; 清华大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
研究针对叙事长视频生成中保持角色身份的挑战，提出SlotMem框架，通过字符语义探测器定位视觉令牌，记忆编码器压缩记忆，记忆写入器更新记忆，逐角色交叉注意力检索记忆，提升了远程角色一致性与视频质量。
AI 中文摘要
在叙事长视频生成中，跨场景转换和长时间间隔保持重复角色身份是一项核心挑战。针对全局一致性的方法在检索记忆时使用的线索与角色身份保存不一致，而最近以角色为中心的变体仍依赖于粗糙的帧级键值记忆，将身份与偶然视觉因素纠缠在一起，且在有限内存容量下缺乏连续更新机制。为解决这些限制，我们提出了SlotMem，一种用于多角色叙事长视频生成的字符可寻址内部存储器框架。具体而言，SlotMem使用字符语义探测器从交叉注意力响应中定位与角色相关的视觉令牌，并使用记忆编码器将DiT令牌压缩为紧凑的按角色插槽记忆。随着生成过程的进行，记忆写入器用新观察结果保守地更新每个角色的记忆，而逐角色交叉注意力检索角色记忆并仅将其注入同一角色的局部令牌中。在多个叙事长视频生成基准上的实验表明，SlotMem在保持可比视频质量的同时，提高了现有基线的远程角色一致性。我们的代码可在该https网址获取。
英文摘要
Maintaining recurring character identities across scene transitions and long temporal gaps is a central challenge in narrative long video generation. Methods targeting global consistency often retrieve memory using cues that are not aligned with character identity preservation, while recent character-centric variants still rely on coarse frame-level kv memory that entangles identity with incidental visual factors and lacks a continuous update mechanism under limited memory capacity. To address these limitations, we propose SlotMem, a character-addressable internal memory framework for multi-character narrative long video generation. Specifically, SlotMem uses a Character-Semantic Probe to localize character-relevant visual tokens from cross-attention responses, and a Memory Encoder to compress DiT tokens into compact role-wise slot memory. As generation proceeds, a Memory Writer conservatively updates each character's memory with new observations, while Character-Wise Cross-Attention retrieves the role memory and injects it only into localized tokens of the same character. Experiments on multiple narrative long video generation benchmarks show that SlotMem improves long-range character consistency over existing baselines, while maintaining comparable video quality. Our code is available at https://github.com/YilaiLiu-HKU/SlotMem.
