OmniRay: camera-agnostic monocular 3D reconstruction with composite ray fields
zhenjun_zhao · x · 2026-10-24
OmniRay is a feed-forward camera-agnostic monocular 3D reconstruction model requiring no intrinsics or camera-type input at inference.
- Motivation: most monocular models assume pinhole cameras and distort on fisheye/panoramic imagery; prior camera-aware models use a single fixed basis for all camera geometries
- The Ray Module composes three components — a monotone radial base, a spherical-harmonic residual, and a free-form residual — as successive rotations on the sphere to accommodate diverse camera geometries
- From a single image it predicts per-pixel ray directions and up-to-scale radial distances to recover 3D point clouds, setting a new state of the art in camera-agnostic monocular 3D reconstruction
Related event: OmniRay: Camera-Agnostic Monocular 3D Reconstruction Without Intrinsics(2 posts)→
More from Research
- NVIDIA and UW Protein Design team up to release RFD4-Proteina for designed biology — AllThingsApx · 2026-10-24
- T²Quant: task- and timestep-aware mixed precision for quantizing diffusion LMs — Ofek Avivi · 2026-10-24
- Berkeley & AWS: agents return faster code with Lean proofs of equivalence, 1.85x mean speedup — xiuyu_l · 2026-10-24
- Researcher: perplexity evals favor code-only models; pass@k would be far more informative — eliebakouch · 2026-10-24
- Google Discover's SDF filters stale articles before ranking — stale reports down 54.9% in two years — gaganghotra_ · 2026-10-24
- Google says Gemini 4 Argon found fusion counterexamples and designed 5 lab-verified enzymes in hours — Eric520CC · 2026-10-24