Research map / 3D vision / 3D generation
EvDiff3D: Event-Aware Diffusion Repair for High-Fidelity Event-Based 3D Reconstruction
Kanghao Chen, Zixin Zhang, Hangyu Li, Lin Wang, Zeyu Wang
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 event-based depth loss (ablation) | 3DGS baseline · Ev-DL3DV | PSNR: 16.78 -> 17.92 | Improved |
| add event-based depth loss (ablation) | 3DGS baseline · Ev-DL3DV | LPIPS: 0.3596 -> 0.3311 | Improved |
| add event-aware diffusion repair (ablation) | EvDiff3D with depth loss · Ev-DL3DV | PSNR: 17.92 -> 18.86 | Improved |
| add event-aware diffusion repair (ablation) | EvDiff3D with depth loss · Ev-DL3DV | LPIPS: 0.3311 -> 0.3243 | Improved |
| add event CLIP feature conditioning (ablation) | event-aware diffusion model · Ev-DL3DV | PSNR: 0.11 PSNR drop when removed | Improved |
| set lambda_repair between 0.5 and 2.0 (ablation) | EvDiff3D · Ev-DL3DV (single sequence) | PSNR | No clear effect |
| increase lambda_repair too large (ablation) | EvDiff3D · Ev-DL3DV (single sequence) | PSNR | Hurt |
| reduce event views to 1/2 of original (ablation) | EvDiff3D · Ev-DL3DV (single sequence) | PSNR: 18.27 dB at 1/2 views | Mixed |
| reduce event views below 1/2 (to 1/4) (ablation) | EvDiff3D · Ev-DL3DV (single sequence) | PSNR | Hurt |
| EvDiff3D versus Event3DGS | Event3DGS · Ev-DL3DV | PSNR: 16.82 -> 18.86 | Improved |
| EvDiff3D versus Event3DGS | Event3DGS · Ev-DL3DV | LPIPS: 0.3584 -> 0.3243 | Improved |
| add diffusion repair stage (training cost) | Event3DGS · Training time | minutes: 15 min -> 40 min | Hurt |
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.
- Event3DGS — Event-3DGS: Event-based 3D Reconstruction Using 3D Gaussian Splatting (NeurIPS 2024) · better on 8 of 9 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
- EvGGS: A Collaborative Learning Framework for Event-based Generalizable Gaussian Splatting (ICML 2024 · similarity 0.88)
- Event-3DGS: Event-based 3D Reconstruction Using 3D Gaussian Splatting (NeurIPS 2024 · similarity 0.88)
- EF-3DGS: Event-Aided Free-Trajectory 3D Gaussian Splatting (NeurIPS 2025 · similarity 0.87)
- AE-NeRF: Augmenting Event-Based Neural Radiance Fields for Non-ideal Conditions and Larger Scenes (AAAI 2025 · similarity 0.86)
- EventSplat: 3D Gaussian Splatting from Moving Event Cameras for Real-time Rendering (CVPR 2025 · similarity 0.86)
- EvFocus: Learning to Reconstruct Sharp Images from Out-of-Focus Event Streams (ICML 2025 · 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.
