Research map / 3D vision / NeRF and radiance fields
Charge: A Comprehensive Novel View Synthesis Benchmark and Dataset to Bind Them All
Michal Nazarczuk, Thomas Tanay, Arthur Moreau, Zhensong Zhang, Eduardo Pérez-Pellitero
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 training views from 3 to 9 | 4DGS · Charge Sparse | PSNR: 19.71 -> 26.67 | Improved |
| increase training views from 3 to 6 | 4DGS · Charge Sparse | PSNR: 19.71 -> 23.93 | Improved |
| evaluate on dynamic regions instead of static regions | 4DGS · Charge Dense | PSNR: 31.85 -> 26.84 | Hurt |
| align target cameras using target views instead of Umeyama on source views | WorldMirror · Charge static Dense NVS | PSNR: 20.10 -> 24.44 | Improved |
| align target cameras using target views instead of Umeyama on source views | WorldMirror · Charge static Sparse-3 NVS | PSNR: 17.54 -> 20.94 | Improved |
| use WorldMirror instead of AnySplat | WorldMirror · Charge static Sparse-3 NVS | PSNR: 12.15 -> 17.54 | Improved |
| use Spline Fast trajectory instead of Random Walk Slow | 4DGS · Charge Mono | PSNR: 23.38 -> 24.85 | Improved |
| use MoSca instead of 4DGS on Random Walk Slow trajectory | MoSca · Charge Mono (Random Walk Slow) | PSNR: 23.38 -> 24.29 | Improved |
| use MoSca instead of D-3DGS on Spline Fast trajectory | MoSca · Charge Mono (Spline Fast) | PSNR: 24.88 -> 23.39 | Hurt |
| use pi3 instead of VGGT for camera pose at Sparse-3 | pi3 · Charge static Sparse-3 | AUC@5: 0.3581 -> 0.1900 | Hurt |
| use pi3 instead of VGGT for depth in Dense | pi3 · Charge static Dense | AbsRel: 0.1085 -> 0.1344 | Mixed |
| use Ex4DGS instead of 4DGS in dense setup | Ex4DGS · Charge Dense | PSNR: 28.94 -> 29.75 | Improved |
Closest papers by meaning
- ViVid-1-to-3: Novel View Synthesis with Video Diffusion Models (CVPR 2024 · similarity 0.87)
- MultiDiff: Consistent Novel View Synthesis from a Single Image (CVPR 2024 · similarity 0.87)
- CAT4D: Create Anything in 4D with Multi-View Video Diffusion Models (CVPR 2025 · similarity 0.87)
- NAMVIS: Next-Scale Autoregressive Multi-View Image Synthesis (NeurIPS 2026 · similarity 0.87)
- GenRec: Knowing Where to Reconstruct and Where to Generate (NeurIPS 2026 · similarity 0.86)
- MVSplat360: Feed-Forward 360 Scene Synthesis from Sparse Views (NeurIPS 2024 · 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.
