Research map / 3D vision / Gaussian splatting
HiReFF: High-Resolution Feedforward Human Reconstruction from Uncalibrated Sparse-View Video
YIMING JIANG, Hanzhang Tu, Wenfeng Song, Siyou Lin, Liang An, Shuai Li, Aimin Hao, Yebin Liu
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 |
|---|---|---|---|
| use high-resolution side-tuning so the backbone stays at 518 while supervising at 2072 (ablation) | HiReFF · Training VRAM | VRAM (MiB): OOM (2K input, no side-tuning) vs 59095 with side-tuning | Improved |
| raise supervision/render resolution from 518 to 2072 with 518 backbone input (ablation) | HiReFF · Training VRAM | VRAM (MiB): 40503 -> 44125 | Hurt |
| feed 2K instead of 0.5K images to backbone (with side-tuning) (ablation) | HiReFF · Inference speed | FPS: 4.40 -> 3.01 FPS | Hurt |
| use 1K instead of 0.5K backbone input when rendering at 2K (ablation) | HiReFF · Inference speed | FPS: 4.40 -> 4.02 FPS | Hurt |
| add Gaussian-wise foreground masking (ablation) | HiReFF · DNA-Rendering qualitative | foreground reconstruction quality | Improved |
| HiReFF vs AnySplat (no camera optimization) | HiReFF · DNA-Rendering test 4 views | PSNR: 23.7844 -> 26.5138 | Improved |
| HiReFF (uncalibrated, 4 views) vs GPS-Gaussian (GT cameras, 8 views) | HiReFF · DNA-Rendering test | SSIM: 0.9172 (GPS-Gaussian) vs 0.9164 | Mixed |
Closest papers by meaning
- Forge4D: Feed-Forward 4D Human Reconstruction and Interpolation from Uncalibrated Sparse-View Videos (ECCV 2026 · similarity 0.89)
- ODHSR: Online Dense 3D Reconstruction of Humans and Scenes from Monocular Videos (CVPR 2025 · similarity 0.87)
- MVSplat360: Feed-Forward 360 Scene Synthesis from Sparse Views (NeurIPS 2024 · similarity 0.87)
- Motion Matters: Compact Gaussian Streaming for Free-Viewpoint Video Reconstruction (NeurIPS 2025 · similarity 0.87)
- Novel View Synthesis from A Few Glimpses via Test-Time Natural Video Completion (NeurIPS 2025 · similarity 0.86)
- HiFi4G: High-Fidelity Human Performance Rendering via Compact Gaussian Splatting (CVPR 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.
