Ask a Question

Prefer a chat interface with context about you and your work?

TexDreamer: Towards Zero-Shot High-Fidelity 3D Human Texture Generation

TexDreamer: Towards Zero-Shot High-Fidelity 3D Human Texture Generation

Texturing 3D humans with semantic UV maps remains a challenge due to the difficulty of acquiring reasonably unfolded UV. Despite recent text-to-3D advancements in supervising multi-view renderings using large text-to-image (T2I) models, issues persist with generation speed, text consistency, and texture quality, resulting in data scarcity among existing datasets. We …