Research map / 3D vision / NeRF and radiance fields
Semantic-Aware Implicit Neural Audio-Driven Video Portrait Generation
Xian Liu, Yinghao Xu, Qianyi Wu, Hang Zhou, Wayne Wu, Bolei Zhou
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 the Torso Deformation module (ablation) | SSP-NeRF · Testset A (full resolution, ablation) | PSNR: 27.472 -> 32.785 | Improved |
| add the Torso Deformation module (ablation) | SSP-NeRF · Testset A (full resolution, ablation) | SyncNet confidence: 4.744 -> 4.993 | Improved |
| add the Torso Deformation module (ablation) | SSP-NeRF · Testset A (full resolution, ablation) | LMD: 4.635 -> 4.495 | Improved |
| feed only head pose, not audio, to the deformation function (ablation) | SSP-NeRF · Testset A (full resolution, ablation) | SyncNet confidence: 3.871 -> 4.993 | Improved |
| feed only head pose, not audio, to the deformation function (ablation) | SSP-NeRF · Testset A (full resolution, ablation) | LMD: 5.329 -> 4.495 | Improved |
| add the semantic parsing branch and semantic loss (ablation) | SSP-NeRF · Testset A (full resolution, ablation) | PSNR: 29.479 -> 32.785 | Improved |
| add the semantic parsing branch and semantic loss (ablation) | SSP-NeRF · Testset A (full resolution, ablation) | SyncNet confidence: 4.562 -> 4.993 | Improved |
| add dynamic semantic-guided ray sampling (ablation) | SSP-NeRF · Testset A (full resolution, ablation) | PSNR: 29.514 -> 32.785 | Improved |
| add dynamic semantic-guided ray sampling (ablation) | SSP-NeRF · Testset A (full resolution, ablation) | SSIM: 0.826 -> 0.876 | Improved |
| add rough 3D facial information via anchored latent codes (ablation) | SSP-NeRF · Testset A (full resolution, ablation) | PSNR: 31.059 -> 32.785 | Improved |
| add rough 3D facial information via anchored latent codes (ablation) | SSP-NeRF · Testset A (full resolution, ablation) | LMD: 4.683 -> 4.495 | Improved |
| use one unified NeRF with semantic sampling and deformation instead of separate head and torso NeRFs | AD-NeRF · Testset A (full resolution 450x450) | PSNR: 29.186 -> 32.785 | Improved |
Closest papers by meaning
- AE-NeRF: Audio Enhanced Neural Radiance Field for Few Shot Talking Head Synthesis (AAAI 2024 · similarity 0.91)
- Teller: Real-Time Streaming Audio-Driven Portrait Animation with Autoregressive Motion Generation (CVPR 2025 · similarity 0.89)
- GoHD: Gaze-oriented and Highly Disentangled Portrait Animation with Rhythmic Poses and Realistic Expressions (AAAI 2025 · similarity 0.88)
- DAWN: Dynamic Frame Avatar with Non-autoregressive Diffusion Framework for Talking head Video Generation (ICLR 2025 · similarity 0.88)
- Hallo3: Highly Dynamic and Realistic Portrait Image Animation with Video Diffusion Transformer (CVPR 2025 · similarity 0.88)
- Real-time 3D-aware Portrait Video Relighting (CVPR 2024 · similarity 0.87)
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.
