Research map / 3D vision / Face and shape
FastGHA: Generalized Few-Shot 3D Gaussian Head Avatars with Real-Time Animation
Xinya Ji, Sebastian Weiss, Manuel Kansy, Jacek Naruniec, Xun Cao, Barbara Solenthaler, Derek Bradley
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 pretrained SD-Turbo VAE weights instead of training encoder/decoder from scratch (ablation) | FastGHA · Ava-256 ablation | PSNR: 20.789 -> 21.274 | Improved |
| add VGGT-derived geometry loss L_geo (ablation) | FastGHA · Ava-256 ablation | PSNR: 21.132 -> 21.274 | Improved |
| add per-Gaussian features (ablation) | FastGHA · Ava-256 ablation | PSNR: 21.053 -> 21.274 | Improved |
| use DINOv3 features instead of Sapiens features (ablation) | FastGHA · Ava-256 ablation | PSNR: 21.081 -> 21.274 | Improved |
| add NeRSemble data to training (Ava-256 only vs both) (ablation) | FastGHA · NeRSemble held-out | PSNR: 22.3 -> 24.0 | Improved |
| add NeRSemble data to training (ablation) | FastGHA · Ava-256 held-out | PSNR: 22.2 -> 22.5 | Improved |
| increase number of input views from 2 to 6 (ablation) | FastGHA · Animation speed | FPS: 128 -> 32 | Hurt |
| replace Avat3r cross-attention animation with lightweight per-Gaussian MLP | FastGHA · Animation speed | FPS: 8 -> 62 | Improved |
Later papers that compared with this method
1 later accepted paper reports FastGHA as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- Large-Scale High-Quality 3D Gaussian Head Reconstruction from Multi-View Captures (ECCV 2026 · better on 1 of 1 shared result)
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- Avat3r — Avat3r: Large Animatable Gaussian Reconstruction Model for High-fidelity 3D Head Avatars (ICCV 2025) · better on 4 of 5 shared results
- GPAvatar — GPAvatar: High-fidelity Head Avatars by Learning Efficient Gaussian Projections (CVPR 2025) · better on 4 of 4 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Gaussian Head Avatar: Ultra High-fidelity Head Avatar via Dynamic Gaussians (CVPR 2024 · similarity 0.92)
- GPAvatar: High-fidelity Head Avatars by Learning Efficient Gaussian Projections (CVPR 2025 · similarity 0.91)
- Generalizable and Animatable Gaussian Head Avatar (NeurIPS 2024 · similarity 0.91)
- HERA: Hybrid Explicit Representation for Ultra-Realistic Head Avatars (CVPR 2025 · similarity 0.91)
- GaussianAvatars: Photorealistic Head Avatars with Rigged 3D Gaussians (CVPR 2024 · similarity 0.90)
- FlashAvatar: High-fidelity Head Avatar with Efficient Gaussian Embedding (CVPR 2024 · similarity 0.90)
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.
