人机共叙的历史逻辑——AIGC历史动画中的框架预设与核校机制

The Historical Logic of Human-AI Co-Narration—Pre-Generation Framework Design and Post-Generation Verification in AIGC Historical Animation

摘要: 本文以热奈特“时态–语式–语态”框架为分析工具，以跨朝代AIGC动画《丝路·镯光》唐代、明代两集及其现存生产材料为探索性个案，比较故事稿、分镜脚本、保留的初始生成片段、核校记录与成片，并选取典型的时序错置与聚焦越界案例进行过程追踪。研究发现，历史时序、聚焦规范和叙述身份可在生成前转化为框架约束，为AI生成执行划定信息与形式边界；框架不能自动排除的史实错置、聚焦越界和媒介连续性中断，仍需生成后的人工核校。本文据此提出“前置框架优先–生成执行受限–后置核校兜底”的控制链条。由于研究未设置对照条件、独立编码及完整生成日志，相关发现应被理解为单一个案中的机制描述，而非框架效果的因果证明。

Abstract: Drawing on Gérard Genette’s categories of tense, mood, and voice, this article presents an exploratory case study of the Tang and Ming Dynasty episodes of the cross-dynastic AIGC animation series The Silk Road Bracelet Light. It compares story drafts, storyboard scripts, initial generated segments and verification records, and final videos. It also selects typical cases of chronological displacement and focalization boundary-crossing for process tracing. The case indicates that historical sequence, focalization norms, and narratorial status can be operationalized as pre-generation constraints, thereby delineating informational and formal boundaries for the AI generation process. However, issues such as historical anachronisms, breaches of focal norms, and disruptions in media continuity—which cannot be automatically precluded by the framing—still require post-generation human verification. Accordingly, this paper proposes a control chain comprising “prioritization of framing constraints,” “constrained generation execution,” and “safeguarding via post-generation verification.” Given the absence of control conditions, independent coding, and complete generation logs in the study, the findings should be interpreted as a description of mechanisms within a single case study rather than as causal proof of the framing’s effects.