Research map / Video / Video generation
EvDiff: High Quality Video with an Event Camera
Weilun Li, Lei Sun, Ruixi Gao, Qi Jiang, Yuqin Ma, Kaiwei Wang, Ming-Hsuan Yang, Luc Van Gool, Danda Paudel
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 Stage 1 DiT training on large image dataset (ablation) | EvDiff · BS-ERGB | MSE: 0.0726 -> 0.0463 | Improved |
| add E2VID-style degradation model (ablation) | EvDiff · BS-ERGB | MSE: 0.0850 -> 0.0463 | Improved |
| add EGM degradation (ablation) | EvDiff · BS-ERGB | FVD: 1683 -> 984 | Improved |
| add SII degradation (ablation) | EvDiff · BS-ERGB | FVD: 1226 -> 984 | Improved |
| add AD degradation (ablation) | EvDiff · BS-ERGB | FVD: 2324 -> 984 | Improved |
| add surrogate distillation stage (ablation) | EvDiff · BS-ERGB | MSE: 0.0660 -> 0.0463 | Improved |
| add Stage 3 joint fine-tuning (ablation) | EvDiff · BS-ERGB | LPIPS: 0.5740 -> 0.4023 | Improved |
| add ETF temporal fusion module (ablation) | EvDiff · BS-ERGB | MSE: 0.0586 -> 0.0463 | Improved |
| use ETF instead of ConvGRU (ablation) | EvDiff · BS-ERGB | MSE: 0.0500 -> 0.0463 | Improved |
| use ETF instead of ConvLSTM (ablation) | EvDiff · DSEC | MSE: 0.0620 -> 0.0476 | Improved |
| use EvEncoder instead of E2VID + VAE-Encoder (ablation) | EvDiff · BS-ERGB | FID: 231 -> 148 | Improved |
| replace multi-step ControlNet with one-step EvDiff | SD3 · 512x512 single frame | FLOPs: 96.56T -> 2.18T | Improved |
Closest papers by meaning
- EvDiff3D: Event-Aware Diffusion Repair for High-Fidelity Event-Based 3D Reconstruction (AAAI 2026 · similarity 0.91)
- Repurposing Pre-trained Video Diffusion Models for Event-based Video Interpolation (CVPR 2025 · similarity 0.89)
- E2HQV: High-Quality Video Generation from Event Camera via Theory-Inspired Model-Aided Deep Learning (AAAI 2024 · similarity 0.89)
- EvFocus: Learning to Reconstruct Sharp Images from Out-of-Focus Event Streams (ICML 2025 · similarity 0.88)
- AE-NeRF: Augmenting Event-Based Neural Radiance Fields for Non-ideal Conditions and Larger Scenes (AAAI 2025 · similarity 0.88)
- LED: A Large-scale Real-world Paired Dataset for Event Camera Denoising (CVPR 2024 · 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.
