Research map / Video / Video generation
High-Fidelity and Freely Controllable Talking Head Video Generation
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| combine learned sparse landmarks with predefined dense face landmarks (ablation) | PECHead · TalkHead-1KH | L1: 0.0446 -> 0.0357 | Helped |
| combine learned sparse landmarks with predefined dense face landmarks (ablation) | PECHead · VFHQ | L1: 0.0491 -> 0.0435 | Helped |
| use both landmark types rather than predefined landmarks alone (ablation) | PECHead · TalkHead-1KH | L1: 0.0426 -> 0.0357 | Helped |
| use both landmark types rather than directly merging them (ablation) | PECHead · TalkHead-1KH | L1: 0.0439 -> 0.0357 | Helped |
| use motion-aware multi-scale feature alignment instead of single-level feature concatenation (ablation) | PECHead · TalkHead-1KH | L1: 0.0430 -> 0.0375 | Helped |
| add the full video framework with context adaptation and propagation to MMFA (ablation) | PECHead · TalkHead-1KH | L1: 0.0375 -> 0.0357 | Helped |
| add the full video framework with context adaptation and propagation to MMFA (ablation) | PECHead · VFHQ | FVD: 209.9 -> 201.6 | Helped |
| use PECHead instead of the strongest listed baseline | same-identity reconstruction · VoxCeleb2 | PSNR: 26.88 -> 26.96 | Helped |
| use PECHead instead of the strongest listed baseline | same-identity reconstruction · TalkHead-1KH | PSNR: 25.53 -> 26.76 | Helped |
| use PECHead instead of the strongest listed baseline | same-identity reconstruction · CelebV-HQ | PSNR: 22.33 -> 24.29 | Helped |
| use PECHead instead of the strongest listed baseline | same-identity reconstruction · VFHQ | PSNR: 22.60 -> 23.03 | Helped |
| use PECHead instead of the strongest listed baseline | cross-identity reenactment · CelebV-HQ | CSIM: 0.716 -> 0.733 | Helped |
Closest papers by meaning
- Synergizing Motion and Appearance: Multi-Scale Compensatory Codebooks for Talking Head Video Generation (CVPR 2025 · similarity 0.92)
- Occlusion-Insensitive Talking Head Video Generation via Facelet Compensation (AAAI 2025 · similarity 0.91)
- AnyTalk: Multi-modal Driven Multi-domain Talking Head Generation (AAAI 2025 · similarity 0.90)
- Diffusion-based Realistic Listening Head Generation via Hybrid Motion Modeling (CVPR 2025 · similarity 0.90)
- GoHD: Gaze-oriented and Highly Disentangled Portrait Animation with Rhythmic Poses and Realistic Expressions (AAAI 2025 · similarity 0.89)
- FG-EmoTalk: Talking Head Video Generation with Fine-Grained Controllable Facial Expressions (AAAI 2024 · similarity 0.89)
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.