Research map / 3D vision / Gaussian splatting
Radiative Gaussian Splatting for Efficient X-ray Novel View Synthesis
Yuanhao Cai, Yixun Liang, Jiahao Wang, Angtian Wang, Yulun Zhang, Xiaokang Yang, Zongwei Zhou, Alan Yuille
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 |
|---|---|---|---|
| Replace SfM initialization with ACUI (ablation) | 3DGS for X-ray NVS · X-ray NVS average | PSNR (dB): 37.213 -> 38.872 | Improved |
| Replace SfM initialization with ACUI (ablation) | 3DGS for X-ray NVS · X-ray NVS average | training time (s): 1898 -> 1172 | Improved |
| Replace RGB SH Gaussian model and rasterization with radiative model and DRR (ablation) | 3DGS + ACUI · X-ray NVS average | PSNR (dB): 38.872 -> 43.404 | Improved |
| Replace RGB SH Gaussian model with radiative model and DRR (ablation) | 3DGS + ACUI · X-ray NVS average | inference speed (fps): 72 -> 148 | Improved |
| Use FDK initialization instead of cuboid ACUI (ablation) | X-Gaussian · X-ray NVS average | PSNR (dB): 43.404 -> 43.470 | No clear effect |
| Use cuboid ACUI instead of random point initialization (ablation) | X-Gaussian · X-ray NVS average | PSNR (dB): 41.329 -> 43.404 | Improved |
| Increase number of radiation features N_f from 1 to 16 (ablation) | X-Gaussian · X-ray NVS average | PSNR (dB): 38.818 -> 43.404 | Improved |
| Increase N_f from 16 to 32 (ablation) | X-Gaussian · X-ray NVS average | PSNR (dB): 43.404 -> 43.417 | No clear effect |
| Use X-Gaussian novel views as extra projections for FDK in sparse-view CT instead of NAF | FDK reconstruction · Sparse-view CT (5 real views) | PSNR (dB): 21.28 -> 22.60 | Improved |
| Use X-Gaussian instead of NAF for novel view synthesis | X-ray NVS · X-ray NVS average | PSNR (dB): 36.942 -> 43.404 | Improved |
| Set init interval d to 8 instead of 1 (ablation) | X-Gaussian ACUI · X-ray NVS average | PSNR (dB): 42.853 -> 43.404 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- NeAT — NeAT: Learning Neural Implicit Surfaces With Arbitrary Topologies From Multi-View Images (CVPR 2023) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- R$^2$-Gaussian: Rectifying Radiative Gaussian Splatting for Tomographic Reconstruction (NeurIPS 2024 · similarity 0.92)
- DDGS-CT: Direction-Disentangled Gaussian Splatting for Realistic Volume Rendering (NeurIPS 2024 · similarity 0.88)
- Structure Consistent Gaussian Splatting with Matching Prior for Few-shot Novel View Synthesis (NeurIPS 2024 · similarity 0.88)
- Volumetrically Consistent 3D Gaussian Rasterization (CVPR 2025 · similarity 0.88)
- COLMAP-Free 3D Gaussian Splatting (CVPR 2024 · similarity 0.88)
- 3DGUT: Enabling Distorted Cameras and Secondary Rays in Gaussian Splatting (CVPR 2025 · similarity 0.87)
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.
