Research map / Video / Video segmentation and tracking
Continuous Space-Time Video Super-Resolution with 3D Fourier Fields
Alexander Becker, Julius Erbach, Dominik Narnhofer, Konrad Schindler
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 |
|---|---|---|---|
| increase number of basis functions from 128 to 256 (ablation) | V3 · Adobe240 (Average) | PSNR: 30.55 -> 31.55 | Improved |
| increase number of basis functions from 256 to 512 (ablation) | V3 · Adobe240 (Average) | PSNR: 31.55 -> 32.16 | Improved |
| increase number of basis functions from 512 to 768 (ablation) | V3 · Adobe240 (Average) | PSNR: 32.16 -> 32.29 | No clear effect |
| learn the basis frequencies instead of fixing a uniform 3D Fourier grid (ablation) | V3 · Adobe240 (Average) | PSNR: 24.80 -> 32.16 | Improved |
| learn the basis frequencies instead of fixing a uniform 3D Fourier grid (ablation) | V3 · Vid4 | PSNR: 22.85 -> 26.72 | Improved |
| train with 100% instead of 50% of iterations (ablation) | V3 · Vid4 x4 | PSNR: 26.71 -> 26.76 | No clear effect |
| train with 100% instead of 25% of iterations (ablation) | V3 · Adobe240 (Average) | PSNR: 31.84 -> 32.16 | Improved |
| scale encoder attention embedding from 90 to 144 (V3-Large) | V3 · Adobe240 (Average) | PSNR: 32.29 -> 32.45 | Improved |
| joint 3D Fourier field instead of decoupled spatial INR plus flow warping | C-STVSR · Adobe240 frame interpolation (S x1 T x8) | PSNR: 29.37 -> 33.43 | Improved |
| joint 3D Fourier field compared with frame-by-frame arbitrary-scale image SR | V3 · REDS val (spatial x4) | PSNR: 28.75 -> 29.92 | Improved |
| joint representation lowers temporal inconsistency | V3 · Vid4 (S x4 T x2) | tOF: 0.323 -> 0.254 | Improved |
| sample with closed-form Gaussian PSF anti-aliasing and joint representation under input noise | V3 · Vid4 with Gaussian noise sigma 5e-2 | PSNR: 24.53 -> 25.18 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.
- BF-STVSR — BF-STVSR: B-Splines and Fourier---Best Friends for High Fidelity Spatial-Temporal Video Super-Resolution (CVPR 2025) · better on 10 of 10 shared results
- MoTIF — MoTIF: Learning Motion Trajectories with Local Implicit Neural Functions for Continuous Space-Time Video Super-Resolution (ICCV 2023) · better on 1 of 1 shared result
- VideoINR — VideoINR: Learning Video Implicit Neural Representation for Continuous Space-Time Super-Resolution (CVPR 2022) · better on 1 of 1 shared result
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
- Time Without Time: Pseudo-Temporal Representation for Space-Time Super-Resolution (CVPR 2026 · similarity 0.90)
- BF-STVSR: B-Splines and Fourier---Best Friends for High Fidelity Spatial-Temporal Video Super-Resolution (CVPR 2025 · similarity 0.90)
- Real-Time Video Super-Resolution With Spatio-Temporal Networks and Motion Compensation (CVPR 2017 · similarity 0.90)
- Spatial-Temporal Space Hand-in-Hand: Spatial-Temporal Video Super-Resolution via Cycle-Projected Mutual Learning (CVPR 2022 · similarity 0.89)
- EvSTVSR: Event Guided Space-Time Video Super-Resolution (AAAI 2025 · similarity 0.87)
- EvEnhancer: Empowering Effectiveness, Efficiency and Generalizability for Continuous Space-Time Video Super-Resolution with Events (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.
