LaVR: Scene Latent Conditioned Generative Video Trajectory Re-Rendering using Large 4D Reconstruction Models
Conference on Computer Vision and Pattern Recognition (CVPR) — 2026
Given a monocular video, the goal of video re-rendering is to generate views of the scene from a novel camera trajectory. Existing methods face two challenges: geometrically unconditioned models lack spatial awareness, leading to drift and deformation under viewpoint changes, while geometrically-conditioned models depend on estimated depth and explicit reconstruction, making them susceptible to depth inaccuracies and calibration errors. We address these challenges by using the implicit geometric knowledge embedded in the latent space of a large 4D reconstruction model to condition the video generation process. These latents capture scene structure in a continuous space without explicit reconstruction, providing a flexible representation that allows the pretrained diffusion prior to regularize errors more effectively. By jointly conditioning on these latents and source camera poses, our model achieves state-of-the-art results on the video re-rendering task.
@article{xie2026lavr,
title={LaVR: Scene Latent Conditioned Generative Video Trajectory Re-Rendering using Large 4D Reconstruction Models},
author={Xie, Mingyang and Khan, Numair and Wang, Tianfu and Dhingra, Naina and Nam, Seonghyeon and Yang, Haitao and Hui, Zhuo and Metzler, Christopher and Vedaldi, Andrea and Pirsiavash, Hamed and others},
journal={arXiv preprint arXiv:2601.14674},
year={2026}
}