CVPR2025
Learning Temporally Consistent Video Depth from Video Diffusion Priors
Jiahao Shao, Yuanbo Yang, Hongyu Zhou, Youmin Zhang, Yujun Shen, Vitor Guizilini, Yue Wang, Matteo Poggi, Yiyi Liao
Abstract
Marigold Input Video Ours Figure 1. We present ChronoDepth, a novel method derived from video diffusion model. It can estimate video depth from arbitrarylength open-world videos while exhibiting high spatial accuracy and state-of-the-art temporal consistency.