Research map / 3D vision / 3D generation
AlignDiff: Learning Physically-Grounded Camera Alignment via Diffusion
Liuyue Xie, Jiancong Guo, Ozan Cakmakci, Andre Araujo, László A. Jeni, Zhiheng Jia
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add line-segment latent conditioning (ablation) | AlignDiff · CO3D (2 views) | angular error: 6.7 -> 4.6 | Improved |
| add line-segment latent conditioning (ablation) | AlignDiff · CO3D (8 views) | angular error: 12.8 -> 8.2 | Improved |
| add localized ray angular loss (ablation) | AlignDiff · CO3D (2 views) | angular error: 5.8 -> 4.6 | Improved |
| add localized ray angular loss (ablation) | AlignDiff · CO3D (8 views) | angular error: 14.7 -> 8.2 | Improved |
| add edge-attention (ablation) | AlignDiff · CO3D (2 views) | angular error: 5.3 -> 4.6 | Improved |
| add optics grounding (ablation) | AlignDiff · CO3D (8 views) | angular error: 11.3 -> 8.2 | Improved |
| use TIPS-g/14 encoder instead of DINOv2-s/14 (ablation) | AlignDiff · CO3D (2 views) | angular error: 8.2 -> 4.6 | Improved |
| use TIPS-g/14 encoder instead of DINOv2-g/14 (ablation) | AlignDiff · CO3D (2 views) | angular error: 4.5 vs 4.6 | No clear effect |
| use AlignDiff instead of RayDiffusion | RayDiffusion · CO3D seen categories (2 views) | angular error: 9.6 -> 3.4 | Improved |
| use AlignDiff instead of RayDiffusion | RayDiffusion · Rectified CO3D seen | rotation accuracy @15 deg (3 views): 92.4 vs 92.3 | No clear effect |
| use AlignDiff instead of RayDiffusion | RayDiffusion · Aria Digital Twins (2 views) | angular error: 28.4 -> 15.4 | Improved |
Closest papers by meaning
- Cameras as Rays: Pose Estimation via Ray Diffusion (ICLR 2024 · similarity 0.88)
- MultiDiff: Consistent Novel View Synthesis from a Single Image (CVPR 2024 · similarity 0.87)
- DiffCalib: Reformulating Monocular Camera Calibration as Diffusion-Based Dense Incident Map Generation (AAAI 2025 · similarity 0.87)
- DiffusionSfM: Predicting Structure and Motion via Ray Origin and Endpoint Diffusion (CVPR 2025 · similarity 0.86)
- UniK3D: Universal Camera Monocular 3D Estimation (CVPR 2025 · similarity 0.86)
- World Tracing: Pixel-Aligned Geometry Beyond the Visible (NeurIPS 2026 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
