Paper Detail
Zhenhao Yang, Xiaoshi Wu, Zhengyao Lv, Xiaoyu Shi, Xintao Wang, Pengfei Wan, Kun Gai, Kwan-Yee K. Wong
Recent advances in video generative models have promoted rapid progress in controllable world models. However, maintaining fine-grained spatio-temporal consistency under long-horizon reasoning remains a key challenge. In this work, we move beyond explicit 3D memory and coarse frame-level implicit modeling, and propose a fine-grained, learnable, and scalable memory for consistent world generation. We first identify two fundamental limitations of naïve learnable memory architectures in long-horizon extrapolation, namely computational inefficiency and attention dispersion. Through a systematic analysis of attention dispersion, we propose DecMem, a decoupled memory architecture that employs Sparse Global Memory for efficient fine-grained access to global history and Anchored Local Memory for stable and high-quality extrapolation. Extensive experiments demonstrate that DecMem significantly outperforms current state-of-the-art methods. By ensuring precise and efficient long-term memory and achieving superior extrapolation capabilities, DecMem enables minute-level controllable long video generation with high fidelity and consistency.
No structured notes yet. Add `summary_sections`, `why_relevant`, `claim_impact`, or `next_action` in `papers.jsonl` to enrich this view.
No ranking explanation is available yet.
No tags.
@misc{yang2026decmem,
title = {DecMem: Towards Minute-Long Consistent World Generation with Decoupled Memory},
author = {Zhenhao Yang and Xiaoshi Wu and Zhengyao Lv and Xiaoyu Shi and Xintao Wang and Pengfei Wan and Kun Gai and Kwan-Yee K. Wong},
year = {2026},
abstract = {Recent advances in video generative models have promoted rapid progress in controllable world models. However, maintaining fine-grained spatio-temporal consistency under long-horizon reasoning remains a key challenge. In this work, we move beyond explicit 3D memory and coarse frame-level implicit modeling, and propose a fine-grained, learnable, and scalable memory for consistent world generation. We first identify two fundamental limitations of naïve learnable memory architectures in long-horizo},
url = {https://huggingface.co/papers/2605.31336},
keywords = {video generative models, world models, spatio-temporal consistency, long-horizon reasoning, learnable memory, attention dispersion, sparse global memory, anchored local memory, video generation, extrapolation, code available, huggingface daily},
eprint = {2605.31336},
archiveprefix = {arXiv},
}
{}