Saved in:
Bibliographic Details
Main Authors: Li, Yaokun, Wang, Shuaixian, Guo, Mantang, Huang, Jiehui, Ding, Taojun, Hu, Mu, Wang, Kaixuan, Shen, Shaojie, Tan, Guang
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2512.03621
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909977345523712
author Li, Yaokun
Wang, Shuaixian
Guo, Mantang
Huang, Jiehui
Ding, Taojun
Hu, Mu
Wang, Kaixuan
Shen, Shaojie
Tan, Guang
author_facet Li, Yaokun
Wang, Shuaixian
Guo, Mantang
Huang, Jiehui
Ding, Taojun
Hu, Mu
Wang, Kaixuan
Shen, Shaojie
Tan, Guang
contents We propose ReCamDriving, a purely vision-based, camera-controlled novel-trajectory video generation framework. While repair-based methods fail to restore complex artifacts and LiDAR-based approaches rely on sparse and incomplete cues, ReCamDriving leverages dense and scene-complete 3DGS renderings for explicit geometric guidance, achieving precise camera-controllable generation. To mitigate overfitting to restoration behaviors when conditioned on 3DGS renderings, ReCamDriving adopts a two-stage training paradigm: the first stage uses camera poses for coarse control, while the second stage incorporates 3DGS renderings for fine-grained viewpoint and geometric guidance. Furthermore, we present a 3DGS-based cross-trajectory data curation strategy to eliminate the train-test gap in camera transformation patterns, enabling scalable multi-trajectory supervision from monocular videos. Based on this strategy, we construct the ParaDrive dataset, containing over 110K parallel-trajectory video pairs. Extensive experiments demonstrate that ReCamDriving achieves state-of-the-art camera controllability and structural consistency.
format Preprint
id arxiv_https___arxiv_org_abs_2512_03621
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle ReCamDriving: LiDAR-Free Camera-Controlled Novel Trajectory Video Generation
Li, Yaokun
Wang, Shuaixian
Guo, Mantang
Huang, Jiehui
Ding, Taojun
Hu, Mu
Wang, Kaixuan
Shen, Shaojie
Tan, Guang
Computer Vision and Pattern Recognition
We propose ReCamDriving, a purely vision-based, camera-controlled novel-trajectory video generation framework. While repair-based methods fail to restore complex artifacts and LiDAR-based approaches rely on sparse and incomplete cues, ReCamDriving leverages dense and scene-complete 3DGS renderings for explicit geometric guidance, achieving precise camera-controllable generation. To mitigate overfitting to restoration behaviors when conditioned on 3DGS renderings, ReCamDriving adopts a two-stage training paradigm: the first stage uses camera poses for coarse control, while the second stage incorporates 3DGS renderings for fine-grained viewpoint and geometric guidance. Furthermore, we present a 3DGS-based cross-trajectory data curation strategy to eliminate the train-test gap in camera transformation patterns, enabling scalable multi-trajectory supervision from monocular videos. Based on this strategy, we construct the ParaDrive dataset, containing over 110K parallel-trajectory video pairs. Extensive experiments demonstrate that ReCamDriving achieves state-of-the-art camera controllability and structural consistency.
title ReCamDriving: LiDAR-Free Camera-Controlled Novel Trajectory Video Generation
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2512.03621