DIAMOND-SSS: Diffusion-Augmented Multi-View Optimization for Data-efficient SubSurface Scattering

2026-01-17

AI Summary

Modeling subsurface scattering (SSS) in neural rendering remains highly challenging due to complex light transport and the reliance on densely captured multi-view, multi-illumination data. This work proposes a data-efficient framework that, for the first time, leverages fine-tuned diffusion models for SSS-aware data augmentation. By integrating illumination-invariant multi-view silhouettes and depth-consistent geometric priors, the method achieves high-fidelity relightable reconstruction under extremely sparse supervision—requiring as few as ten real images. It drastically reduces the need for real-world data capture, enabling up to 95% of training images to be replaced by synthetic counterparts while maintaining state-of-the-art relightable Gaussian rendering quality across all sparsity levels. Compared to SSS-3DGS, the approach reduces real data acquisition by up to 90%.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Large Multimodal Models (LMMs)
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding

Application Category

• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Search and Retrieval-Augmented AI: Efficiency and scalability of Web search engines
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs

Abstract

Subsurface scattering (SSS) gives translucent materials -- such as wax, jade, marble, and skin -- their characteristic soft shadows, color bleeding, and diffuse glow. Modeling these effects in neural rendering remains challenging due to complex light transport and the need for densely captured multi-view, multi-light datasets (often more than 100 views and 112 OLATs). We present DIAMOND-SSS, a data-efficient framework for high-fidelity translucent reconstruction from extremely sparse supervision -- even as few as ten images. We fine-tune diffusion models for novel-view synthesis and relighting, conditioned on estimated geometry and trained on less than 7 percent of the dataset, producing photorealistic augmentations that can replace up to 95 percent of missing captures. To stabilize reconstruction under sparse or synthetic supervision, we introduce illumination-independent geometric priors: a multi-view silhouette consistency loss and a multi-view depth consistency loss. Across all sparsity regimes, DIAMOND-SSS achieves state-of-the-art quality in relightable Gaussian rendering, reducing real capture requirements by up to 90 percent compared to SSS-3DGS.

Problem

• Subsurface Scattering
• Data-efficient Reconstruction
• Neural Rendering
• Sparse Supervision
• Translucent Materials

Innovation

• diffusion-augmented rendering
• subsurface scattering
• data-efficient reconstruction
• multi-view consistency
• relightable Gaussian rendering