Neural LightRig: Unlocking Accurate Object Normal and Material Estimation with Multi-Light Diffusion
This work addresses the challenge of intrinsic estimation for computer vision applications, offering improved accuracy for tasks like relighting, but it is incremental as it builds on existing diffusion and G-buffer methods.
The paper tackles the under-constrained problem of recovering object geometry and materials from a single image by using a multi-light diffusion framework to generate consistent images under varied lighting, resulting in significant outperformance over state-of-the-art methods for accurate surface normal and PBR material estimation.
Recovering the geometry and materials of objects from a single image is challenging due to its under-constrained nature. In this paper, we present Neural LightRig, a novel framework that boosts intrinsic estimation by leveraging auxiliary multi-lighting conditions from 2D diffusion priors. Specifically, 1) we first leverage illumination priors from large-scale diffusion models to build our multi-light diffusion model on a synthetic relighting dataset with dedicated designs. This diffusion model generates multiple consistent images, each illuminated by point light sources in different directions. 2) By using these varied lighting images to reduce estimation uncertainty, we train a large G-buffer model with a U-Net backbone to accurately predict surface normals and materials. Extensive experiments validate that our approach significantly outperforms state-of-the-art methods, enabling accurate surface normal and PBR material estimation with vivid relighting effects. Code and dataset are available on our project page at https://projects.zxhezexin.com/neural-lightrig.