Research map / 3D vision / 3D generation
Volumetric performance capture from minimal camera viewpoints
Andrew Gilbert, Marco Volino, John Collomosse, Adrian Hilton
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 |
|---|---|---|---|
| apply the autoencoder to a 2-camera PVH | probabilistic visual hull · TotalCapture, 2 cameras | MSE x 1e-3: 24.6 -> 7.71 | Improved |
| apply the autoencoder to a 4-camera PVH | probabilistic visual hull · TotalCapture, 4 cameras | MSE x 1e-3: 11.6 -> 7.60 | Improved |
| use overlapping patches (overlap 16) instead of no overlap (32) (ablation) | autoencoder, 2 cameras · TotalCapture 32^3 patches | MSE x 1e-3: 9.20 -> 7.71 | Improved |
| use overlapping patches (overlap 8) instead of no overlap (32) (ablation) | autoencoder, 4 cameras · TotalCapture 32^3 patches | MSE x 1e-3: 8.28 -> 7.33 | Improved |
| reduce patch overlap from 8 to 16 (ablation) | autoencoder, 2 cameras · TotalCapture 32^3 patches | MSE x 1e-3: 7.75 -> 7.71 | No clear effect |
| process the whole 64^3 volume instead of 32^3 patches with overlap 8 (ablation) | autoencoder, 2 cameras · TotalCapture 64^3 PVH | MSE x 1e-3: 6.10 -> 6.00 | No clear effect |
| remove patch overlap for 32^3 patches (overlap 32 vs 8) (ablation) | autoencoder, 2 cameras · TotalCapture 64^3 PVH | MSE x 1e-3: 7.20 -> 6.10 | Improved |
| use 4 cameras in 180 degree arc instead of 2 neighbouring cameras | autoencoder · TotalCapture | MSE x 1e-3: 7.71 -> 7.60 | Improved |
| transfer the TotalCapture model to Human3.6M without retraining | autoencoder, 2 cameras · Human3.6M | MSE x 1e-3: 17.4 -> 12.3 | Improved |
| use the 2-view enhanced PVH instead of an 8-view PVH for 4D reconstruction | Casas et al. pipeline · 4D reconstruction | PSNR: 35.1 -> 34.9 | No clear effect |
| use the 2-view enhanced PVH instead of an 8-view PVH for 4D reconstruction | Casas et al. pipeline · 4D reconstruction | SSIM: 0.898 -> 0.899 | No clear effect |
| use the 2-view enhanced PVH instead of an 8-view PVH (user preference) | Casas et al. pipeline · AMT perceptual study | preference (%): 49.2 -> 50.7 | No clear effect |
Closest papers by meaning
- Deep Volumetric Video From Very Sparse Multi-View Performance Capture (ECCV 2018 · similarity 0.90)
- A Deeper Look into DeepCap (TPAMI 2021 · similarity 0.89)
- DeepCap: Monocular Human Performance Capture Using Weak Supervision (CVPR 2020 · similarity 0.88)
- Feed-forward Human Performance Capture via Progressive Canonical Space Updates (ICLR 2026 · similarity 0.86)
- RePerformer: Immersive Human-centric Volumetric Videos from Playback to Photoreal Reperformance (CVPR 2025 · similarity 0.84)
- ImViD: Immersive Volumetric Videos for Enhanced VR Engagement (CVPR 2025 · similarity 0.83)
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.
