2, the authors introduce a local modality mixer that fuses patch-aligned RGB, target-mask, and pointmap features before cross-view reasoning, separating local cross-modal association from global view aggregation.
Generative and agentic pipeline reconstructs scene-aligned 3D objects from casual photos
GATOR fuses patch-aligned RGB, mask, and pointmap features with text guidance, then refines generated assets through an observation-guided edit-render-review loop.
Big Tech
Qirui Wu · Stan Birchfield · Hesam Rabeti · Angel X. Chang · Bowen Wen
NVIDIA · Simon Fraser University
Research Digest··2 min read
GATOR reconstructs complete, textured 3D objects and their scene-relative pose from one or a few casual images.
Why this paper
From NVIDIA and Simon Fraser University
In one line
GATOR reconstructs complete, scene-aligned 3D objects with pose from one or more casual images using generative and agentic methods.
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.
§