# DCARL: A Divide-and-Conquer Framework for Autoregressive Long-Trajectory Video Generation

Source: [DCARL: A Divide-and-Conquer Framework for Autoregressive Long-Trajectory Video Generation](<https://www.sota2.com/research/paper/69c60ee50636e5bf4afc8386>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-03-28T05:00:04.914Z

Authors: Junyi Ouyang, Wenbin Teng, Gonglin Chen, Yajie Zhao, Haiwei Chen

Published: 2026-03-25T22:07:09.000Z

## Abstract

Long-trajectory video generation is a crucial yet challenging task for world modeling primarily due to the limited scalability of existing video diffusion models (VDMs). Autoregressive models, while offering infinite rollout, suffer from visual drift and poor controllability. To address these issues, we propose DCARL, a novel divide-and-conquer, autoregressive framework that effectively combines the structural stability of the divide-and-conquer scheme with the high-fidelity generation of VDMs. Our approach first employs a dedicated Keyframe Generator trained without temporal compression to establish long-range, globally consistent structural anchors. Subsequently, an Interpolation Generator synthesizes the dense frames in an autoregressive manner with overlapping segments, utilizing the keyframes for global context and a single clean preceding frame for local coherence. Trained on a large-scale internet long trajectory video dataset, our method achieves superior performance in both visual quality (lower FID and FVD) and camera adherence (lower ATE and ARE) compared to state-of-the-art autoregressive and divide-and-conquer baselines, demonstrating stable and high-fidelity generation for long trajectory videos up to 32 seconds in length.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Video Generation on nuScenes 16-second videos (test)](<https://www.sota2.com/research/sota/video-generation-on-nuscenes-16-second-videos-test.md>): Video Generation; nuScenes 16-second videos (test); Overall FID: 19.6.

- [Video Generation on ODV-YT 32-second videos](<https://www.sota2.com/research/sota/video-generation-on-odv-yt-32-second-videos.md>): Video Generation; ODV-YT 32-second videos; FID (Overall): 19.2.

- [Video Generation on DL3DV-10K (test)](<https://www.sota2.com/research/sota/video-generation-on-dl3dv-10k-test.md>): Video Generation; DL3DV 10K (test); Overall FID: 22.8.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2603.24835>)

[PDF](<https://arxiv.org/pdf/2603.24835.pdf>)
