We present MotionCrafter, a framework that leverages video generators to jointly reconstruct 4D geometry and estimate dense motion from a monocular video. The key idea is a joint representation of dense 3D point maps and 3D scene flows in a shared coordinate system, together with a 4D VAE tailored to learn this representation effectively. Unlike prior work that strictly aligns 3D values and latents with RGB VAE latents-despite their fundamentally different distributions-we show that such alignment is unnecessary and can hurt performance. Instead, we propose a new data normalization and VAE training strategy that better transfers diffusion priors and greatly improves reconstruction quality. Extensive experiments on multiple datasets show that MotionCrafter achieves state-of-the-art performance in both geometry reconstruction and dense scene flow estimation, delivering 38.64% and 25.0% improvements in geometry and motion reconstruction, respectively, all without any post-optimization. Project page: https://ruijiezhu94.github.io/MotionCrafter_Page
@article{arxiv.2602.08961,
title = {MotionCrafter: Dense Geometry and Motion Reconstruction with a 4D VAE},
author = {Ruijie Zhu and Jiahao Lu and Wenbo Hu and Xiaoguang Han and Jianfei Cai and Ying Shan and Chuanxia Zheng},
journal= {arXiv preprint arXiv:2602.08961},
year = {2026}
}