Research map / 3D vision / Face and shape
Learning Dynamic Facial Radiance Fields for Few-Shot Talking Head Synthesis
Shuai Shen, Wanhua Li, Zheng Zhu, Yueqi Duan, Jie Zhou, Jiwen Lu
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 |
|---|---|---|---|
| add audio-conditioned differentiable face warping (ablation) | DFRF · Test Set A, 15s clip, 50k iters | PSNR: 29.50 -> 29.66 | Improved |
| add differentiable face warping (ablation) | DFRF · Test Set A | LPIPS: 0.057 -> 0.053 | Improved |
| add differentiable face warping (ablation) | DFRF · Test Set A | SyncNet offset/confidence: -1/4.152 -> 0/4.822 | Improved |
| add differentiable face warping (ablation) | DFRF · Test Set B | SyncNet offset/confidence: -2/2.852 -> 0/4.183 | Improved |
| add differentiable face warping (ablation) | DFRF · Test Set B | SSIM: 0.899 -> 0.899 | No clear effect |
| increase reference images from 1 to 4 (ablation) | DFRF · 15s clip, 10k iters | PSNR: 31.03 -> 31.23 | Improved |
| increase reference images from 4 to 6 (ablation) | DFRF · 15s clip, 10k iters | LPIPS: 0.019 -> 0.020 | No clear effect |
| use DFRF with pretrained base model instead of AD-NeRF | AD-NeRF · 10s clip, 50k iters | SyncNet confidence: 1.313 -> 3.447 | Improved |
| use DFRF instead of AD-NeRF | AD-NeRF · 10s clip, 50k iters | LPIPS: 0.039 -> 0.036 | Improved |
| use DFRF instead of AD-NeRF on PSNR | AD-NeRF · 15s clip, 50k iters | PSNR: 31.32 -> 30.75 | Hurt |
| use DFRF instead of AD-NeRF at 1k iterations | AD-NeRF · Test Set A, 15s clip | LPIPS: 0.084 -> 0.040 | Improved |
| use DFRF instead of AD-NeRF at 40k iterations | AD-NeRF · Test Set A, 15s clip | LPIPS: 0.039 -> 0.037 | Improved |
Closest papers by meaning
- AE-NeRF: Audio Enhanced Neural Radiance Field for Few Shot Talking Head Synthesis (AAAI 2024 · similarity 0.93)
- MimicTalk: Mimicking a personalized and expressive 3D talking face in minutes (NeurIPS 2024 · similarity 0.89)
- Real-time 3D-aware Portrait Video Relighting (CVPR 2024 · similarity 0.89)
- High-Fidelity 3D Head Avatars Reconstruction through Spatially-Varying Expression Conditioned Neural Radiance Field (AAAI 2024 · similarity 0.88)
- DAWN: Dynamic Frame Avatar with Non-autoregressive Diffusion Framework for Talking head Video Generation (ICLR 2025 · similarity 0.88)
- Learning Dynamic Tetrahedra for High-Quality Talking Head Synthesis (CVPR 2024 · similarity 0.88)
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.
