Research map / 3D vision / Face and shape
Feed-Forward One-Shot Animatable Textured Mesh Avatar Reconstruction
Yisheng He
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 |
|---|---|---|---|
| replace per-vertex colors with a UV texture map (ablation) | MeshLAM · VFHQ test | PSNR: 18.09 -> 25.23 | Improved |
| add iterative GRU decoding instead of direct regression (ablation) | MeshLAM · VFHQ test | PSNR: 23.08 -> 25.23 | Improved |
| add texture-space reprojection and unwrapping (ablation) | MeshLAM · VFHQ test | FID: 29.428 -> 22.699 | Improved |
| add part-aware deformation clipping (ablation) | MeshLAM · VFHQ test | FID: 32.405 -> 22.699 | Improved |
| add optional UNet neural renderer (ablation) | MeshLAM · VFHQ test | PSNR: 23.180 -> 25.233 | Improved |
| use 2 GRU iterations instead of 1 (ablation) | MeshLAM · VFHQ test | PSNR: 23.10 -> 25.23 | Improved |
| use 3 GRU iterations instead of 2 (ablation) | MeshLAM · VFHQ test | mixed PSNR/LPIPS/FID: PSNR 25.23 -> 25.38; FID 22.699 -> 23.431 | Mixed |
| initialize Gaussian LAM with MeshLAM mesh instead of FLAME | LAM · VFHQ test | LPIPS: 0.077 -> 0.050 | Improved |
| mesh representation (8K vertices) instead of Gaussians | MeshLAM vs LAM · reconstruction speed | seconds: 5.8 -> 0.7 | Improved |
Closest papers by meaning
- LeGO: Leveraging a Surface Deformation Network for Animatable Stylized Face Generation with One Example (CVPR 2024 · similarity 0.87)
- HERA: Hybrid Explicit Representation for Ultra-Realistic Head Avatars (CVPR 2025 · similarity 0.86)
- GALA: Generating Animatable Layered Assets from a Single Scan (CVPR 2024 · similarity 0.86)
- FATE: Full-head Gaussian Avatar with Textural Editing from Monocular Video (CVPR 2025 · similarity 0.86)
- MeGA: Hybrid Mesh-Gaussian Head Avatar for High-Fidelity Rendering and Head Editing (CVPR 2025 · similarity 0.86)
- Efficient 3D Implicit Head Avatar with Mesh-anchored Hash Table Blendshapes (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.
