The authors start with known geometry, camera poses and a non-overlapping UV parameterization.
Texture-space diffusion generates view-consistent materials for 3D objects
The authors adapt a pretrained video diffusion transformer to reconstruct and generate detailed PBR material maps directly on a shared UV texture atlas.
Big Tech
Jacob Munkberg · Peter Kocsis · Jon Hasselgren
NVIDIA
Research Digest··3 min read
Munkberg, Kocsis and Hasselgren move material generation from rendered views into the two-dimensional texture coordinates attached to a 3D object.
Why this paper
From NVIDIA
In one line
Texture space video diffusion generates high-quality 3D materials from text or photos, avoiding view consistency issues and scaling to 8K.
What we could check
- ·No code link found
- ·No weights link found
- ·No dataset link found
- ·No compute details found
- ·No stated limitations found
- ·No benchmark numbers found
Observed from the paper text and links we have. Absence here means we did not find it, not that it does not exist.
§