REEL：为任意长时程多镜头视频生成学习检索、评估与演化视觉记忆

View PDF HTML (experimental)

Abstract:Memory-based long-video generation aims to combine user control with long-range visual consistency. We organize this process into three functional components: a planner, an adapter, and a generator. Representative multi-shot systems couple memory to generator-specific interfaces or use predefined agent workflows, rather than independently training a memory policy for cross-shot objectives. We introduce REEL, a general-purpose visual-memory planner. Through a unified tool interface, REEL learns to retrieve references, create missing visual evidence, and update persistent memory. Backend-specific adapters translate its evidence plans into supported conditioning inputs, allowing the same policy to guide different black-box video generators. To train REEL, we construct two high-quality datasets, REEL-SFT-4k and REEL-RL-1k, covering diverse interaction patterns and prompt types. Training proceeds in three stages: supervised fine-tuning, single-shot Group Relative Policy Optimization (GRPO), and multi-shot GRPO. Multi-shot GRPO carries policy-generated memory across consecutive shots, so later outcomes guide earlier reference selection and memory updates. REEL-Bench tests identity preservation and intended state changes under text-only requests and timed reference-image uploads. On JoyAI-Echo~1.5 and MiniMax-H3, REEL improves cross-shot consistency on ST-Bench by 15.1\% and 20.7\%, and Core-100 on our proposed REEL-Bench by 2.41 and 8.32 points, respectively, over the no-memory baseline.

Comments:	this https URL
Subjects:	Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
Cite as:	arXiv:2610.16632 [cs.CV]
	(or arXiv:2610.16632v1 [cs.CV] for this version)
	https://doi.org/10.48550/arXiv.2610.16632 arXiv-issued DOI via DataCite (pending registration)

Submission history

From: Hang Xu [view email]
[v1] Mon, 12 Oct 2026 11:54:14 UTC (17,984 KB)