We propose DiffuStereo, a novel system using only sparse cameras (8 in this work) for high-quality 3D human reconstruction. At its core is a novel diffusion-based stereo module, which introduces diffusion models, a type of powerful generative models, into the iterative stereo matching network. To this end, we design a new diffusion kernel and additional stereo constraints to facilitate stereo matching and depth estimation in the network. We further present a multi-level stereo network architecture to handle high-resolution (up to 4k) inputs without requiring unaffordable memory footprint. Given a set of sparse-view color images of a human, the proposed multi-level diffusion-based stereo network can produce highly accurate depth maps, which are then converted into a high-quality 3D human model through an efficient multi-view fusion strategy. Overall, our method enables automatic reconstruction of human models with quality on par to high-end dense-view camera rigs, and this is achieved using a much more light-weight hardware setup. Experiments show that our method outperforms state-of-the-art methods by a large margin both qualitatively and quantitatively.

[ArXiv] [Paper] [Supp] [Code] [Data]

Fig 2. Overview of the DiffuStereo system. Our system consists of three key steps to reconstruct high-quality human models from sparse-view inputs: i) An initial human mesh is predicted by DoubleField, and rendered as the coarse disparity flow; ii) The coarse disparity maps are refined in the diffusion-based stereo to obtain the high-quality depth maps; iii) The initial human mesh and high-quality depth maps are fused as the final high-quality human mesh.

Fig 3. Illustration of the forward process and the reverse process in our diffusion-based stereo.