Reproduce “Advances in Neural 3D Mesh Texturing: A Survey” 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
Sai Raj Kishore Perla, Hao Zhang, Ali Mahdavi-Amiri
This survey paper provides a comprehensive overview of recent advancements in neural 3D mesh texturing, organizing diverse methods for texture synthesis, transfer, and completion into a coherent taxonomy.
Key insights
- Neural 3D mesh texturing has evolved from early methods relying on weak 2D supervision to leveraging powerful pre-trained 2D generative priors like vision-language models and large-scale diffusion models.
- Differentiable rendering is a fundamental enabler, bridging 2D image-space supervision with 3D mesh properties to facilitate gradient-based texture optimization and ensure multi-view consistency.
- There is a growing trend towards generating physically based rendering (PBR) material maps for improved realism and editability, with specialized techniques emerging for domains like 3D human texturization.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.