Reproduce “LDM: Large Tensorial SDF Model for Textured Mesh Generation” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Rengan Xie, Wenting Zheng, Kai Huang, Yizheng Chen, Qi Wang, Qi Ye, Wei Chen, Yuchi Huo
LDM, developed by researchers at Zhejiang University and collaborators, introduces a feed-forward framework that generates high-fidelity, illumination-decoupled textured meshes from single images or text prompts in seconds. It produces smoother geometries and textures that enable flexible relighting and material editing for downstream 3D applications.
Key insights
- Tensorial SDFs, combined with a decoupled color field, effectively represent both geometry and illumination-independent textures, enabling high-fidelity and reusable 3D assets.
- An adaptive, optimizable parameter for SDF-to-density conversion enhances training stability and geometric quality for diverse objects in a generative setting.
- A two-stage training strategy, combining initial volume rendering with subsequent differentiable mesh extraction, allows for learning global structure and refining local details.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.