DCDM: Divide-and-Conquer Diffusion Models for Consistency-Preserving Video Generation

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Zhao, Haoyu, Zhang, Yuang, Cheng, Junqi, Gu, Jiaxi, Lu, Zenghui, Shu, Peng, Wu, Zuxuan, Jiang, Yu-Gang
Format: Preprint
Veröffentlicht: 2026
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866910022213042176
author Zhao, Haoyu
Zhang, Yuang
Cheng, Junqi
Gu, Jiaxi
Lu, Zenghui
Shu, Peng
Wu, Zuxuan
Jiang, Yu-Gang
author_facet Zhao, Haoyu
Zhang, Yuang
Cheng, Junqi
Gu, Jiaxi
Lu, Zenghui
Shu, Peng
Wu, Zuxuan
Jiang, Yu-Gang
contents Recent video generative models have demonstrated impressive visual fidelity, yet they often struggle with semantic, geometric, and identity consistency. In this paper, we propose a system-level framework, termed the Divide-and-Conquer Diffusion Model (DCDM), to address three key challenges: (1) intra-clip world knowledge consistency, (2) inter-clip camera consistency, and (3) inter-shot element consistency. DCDM decomposes video consistency modeling under these scenarios into three dedicated components while sharing a unified video generation backbone. For intra-clip consistency, DCDM leverages a large language model to parse input prompts into structured semantic representations, which are subsequently translated into coherent video content by a diffusion transformer. For inter-clip camera consistency, we propose a temporal camera representation in the noise space that enables precise and stable camera motion control, along with a text-to-image initialization mechanism to further enhance controllability. For inter-shot consistency, DCDM adopts a holistic scene generation paradigm with windowed cross-attention and sparse inter-shot self-attention, ensuring long-range narrative coherence while maintaining computational efficiency. We validate our framework on the test set of the CVM Competition at AAAI'26, and the results demonstrate that the proposed strategies effectively address these challenges.
format Preprint
id arxiv_https___arxiv_org_abs_2602_13637
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle DCDM: Divide-and-Conquer Diffusion Models for Consistency-Preserving Video Generation
Zhao, Haoyu
Zhang, Yuang
Cheng, Junqi
Gu, Jiaxi
Lu, Zenghui
Shu, Peng
Wu, Zuxuan
Jiang, Yu-Gang
Computer Vision and Pattern Recognition
Recent video generative models have demonstrated impressive visual fidelity, yet they often struggle with semantic, geometric, and identity consistency. In this paper, we propose a system-level framework, termed the Divide-and-Conquer Diffusion Model (DCDM), to address three key challenges: (1) intra-clip world knowledge consistency, (2) inter-clip camera consistency, and (3) inter-shot element consistency. DCDM decomposes video consistency modeling under these scenarios into three dedicated components while sharing a unified video generation backbone. For intra-clip consistency, DCDM leverages a large language model to parse input prompts into structured semantic representations, which are subsequently translated into coherent video content by a diffusion transformer. For inter-clip camera consistency, we propose a temporal camera representation in the noise space that enables precise and stable camera motion control, along with a text-to-image initialization mechanism to further enhance controllability. For inter-shot consistency, DCDM adopts a holistic scene generation paradigm with windowed cross-attention and sparse inter-shot self-attention, ensuring long-range narrative coherence while maintaining computational efficiency. We validate our framework on the test set of the CVM Competition at AAAI'26, and the results demonstrate that the proposed strategies effectively address these challenges.
title DCDM: Divide-and-Conquer Diffusion Models for Consistency-Preserving Video Generation
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2602.13637