Research map / 3D vision / Face and shape
UV Volumes for Real-Time Rendering of Editable Free-View Human Performance
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 smooth 3D UV volumes with detailed 2D neural texture stacks | UV Volumes · CMU Panoptic, ZJU Mocap, and H36M novel pose synthesis | LPIPS: 0.146 -> 0.073; 0.120 -> 0.104; 0.170 -> 0.141 | Helped |
| Add the DensePose warm-start losses (ablation) | UV Volumes · CMU Panoptic p1 novel view | LPIPS: 0.060 -> 0.036 | Helped |
| Add perceptual loss (ablation) | UV Volumes · CMU Panoptic p1 novel view | LPIPS: 0.055 -> 0.036 | Helped |
| Add silhouette loss (ablation) | UV Volumes · CMU Panoptic p1 novel view | PSNR: 17.47 -> 30.38 | Helped |
| Use CNN-based Spatial NTS instead of Global NTS (ablation) | UV Volumes · CMU Panoptic p1 novel view | LPIPS: 0.062 -> 0.036 | Helped |
| Use CNN-based Spatial NTS instead of Local NTS (ablation) | UV Volumes · CMU Panoptic p1 novel view | LPIPS: 0.061 -> 0.036 | Helped |
| Use CNN-based Spatial NTS instead of Hyper NTS (ablation) | UV Volumes · CMU Panoptic p1 novel view | LPIPS: 0.059 -> 0.036 | Helped |
| Add ray direction noise (ablation) | UV Volumes · CMU Panoptic p1 novel view | LPIPS: 0.058 -> 0.036 | Helped |
| Use 20 rather than 4 training views (ablation) | UV Volumes · ZJU Mocap performer 313 novel view synthesis | PSNR: 29.56 -> 31.12 | Helped |
| Train with longer video sequences (ablation) | UV Volumes · ZJU Mocap performer 313, 4 training views | PSNR: 29.56 -> 28.96 | Hurt |
| Use UV Volumes for rendering rather than NeuralBody | Human novel-view rendering · ZJU Mocap performer 313 | FPS: 1.5FPS -> 51FPS | Helped |
Closest papers by meaning
- Real-time Free-view Human Rendering from Sparse-view RGB Videos using Double Unprojected Textures (CVPR 2025 · similarity 0.87)
- Human Gaussian Splatting: Real-time Rendering of Animatable Avatars (CVPR 2024 · similarity 0.86)
- RePerformer: Immersive Human-centric Volumetric Videos from Playback to Photoreal Reperformance (CVPR 2025 · similarity 0.86)
- Holoported Characters: Real-time Free-viewpoint Rendering of Humans from Sparse RGB Cameras (CVPR 2024 · similarity 0.86)
- NECA: Neural Customizable Human Avatar (CVPR 2024 · similarity 0.86)
- IntrinsicAvatar: Physically Based Inverse Rendering of Dynamic Humans from Monocular Videos via Explicit Ray Tracing (CVPR 2024 · similarity 0.85)
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.