MVDD: Multi-View Depth Diffusion Models

Zhen Wang1,2, Qiangeng Xu1, Feitong Tan1, Menglei Chai1, Shichen Liu1, Rohit Pandey1, Sean Fanello1 Achuta Kadambi2 Yinda Zhang1
1Google, 2UCLA
Paper arXiv Code (Coming)

MVDD is versatile and can be used for shape generation, shape completion and shape regularization.

Abstract

Denoising diffusion models have demonstrated outstanding results in 2D image generation, yet it remains a challenge to replicate its success in 3D shape generation. In this paper, we propose leveraging multi-view depth, which represents complex 3D shapes in a 2D data format that is easy to denoise. We pair this representation with a diffusion model, MVDD, that is capable of generating high-quality dense point clouds with 20K+ points with fine-grained de- tails. To enforce 3D consistency in multi-view depth, we introduce an epipolar line segment attention that conditions the denoising step for a view on its neighboring views. Additionally, a depth fusion module is incorporated into diffusion steps to further ensure the alignment of depth maps. When augmented with surface reconstruction, MVDD can also produce high-quality 3D meshes. Furthermore, MVDD stands out in other tasks such as depth completion, and can serve as a 3D prior, significantly boosting many downstream tasks, such as GAN inversion. State-of-the-art results from extensive experiments demonstrate MVDD’s excellent ability in 3D shape generation, depth completion, and its potential as a 3D prior for downstream tasks.

Method

To enforce multi- view 3D consistency, MVDD denoises each depth map with an efficient epipolar “line segment” attention. Specifically, by leveraging the denoised value from the current step, MVDD only needs to attend to features on a line segment centered around the back-projected depth (the red dot), rather than the entire epipolar line. To further align the denoised multi-view depths, depth fusion is incorporated after the U-Net in a denoising step.

Shape Generation

MVDD produces high-quality details:

This is an animated gif image, but it does not move This is an animated gif image, but it does not move This is an animated gif image, but it does not move This is an animated gif image, but it does not move This is an animated gif image, but it does not move This is an animated gif image, but it does not move This is an animated gif image, but it does not move This is an animated gif image, but it does not move This is an animated gif image, but it does not move This is an animated gif image, but it does not move This is an animated gif image, but it does not move This is an animated gif image, but it does not move This is an animated gif image, but it does not move This is an animated gif image, but it does not move This is an animated gif image, but it does not move


Our model augmented with surface reconstruction method SAP



Top: PVD (Zhou et al.); Bottom: LION (Zeng et al.)

This is an animated gif image, but it does not move This is an animated gif image, but it does not move This is an animated gif image, but it does not move
This is an animated gif image, but it does not move This is an animated gif image, but it does not move This is an animated gif image, but it does not move


Shape Completion


Shape Regularization



BibTeX

@article{wang2023mvdd,
  author    = {Zhen Wang, Qiangeng Xu, Feitong Tan, Menglei Chai, Shichen Liu, Rohit Pandey, Sean Fanello, Achuta Kadambi, Yinda Zhang},
  title     = {MVDD: Multi-View Depth Diffusion Models},
  journal   = {arXiv},
  year      = {2023},
}