Research map / 3D vision / Depth and stereo
FG-Portrait: 3D Flow Guided Editable Portrait Animation
Yating Xu, Yunqi Miao, Evangelos Ververas, Jiankang Deng, Jifei Song
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 driving landmark condition with 3D flow encoding (ablation) | FG-Portrait ControlNet · VFHQ self-reenactment | APD: 4.001 -> 2.682 | Improved |
| replace predicted flow map with 3D flow (ablation) | FG-Portrait ControlNet · VFHQ cross-reenactment | APD: 12.430 -> 7.764 | Improved |
| add depth-guided sampling (ablation) | 3D flow encoding · VFHQ self-reenactment | APD: 9.659 -> 2.682 | Improved |
| add depth-guided sampling (ablation) | 3D flow encoding · VFHQ self-reenactment | LPIPS: 0.213 -> 0.158 | Improved |
| add depth-guided sampling (ablation) | 3D flow encoding · VFHQ self-reenactment | CSIM: 0.770 -> 0.807 | Improved |
| use 3D flow motion condition instead of other diffusion conditions | FG-Portrait · VFHQ self-reenactment | APD: 3.440 -> 2.682 (vs HunyuanPortrait) | Improved |
| use FG-Portrait instead of baselines on identity preservation | FG-Portrait · VFHQ cross-reenactment | CSIM: 0.484 -> 0.462 (vs Follow-Your-Emoji) | Hurt |
| use FG-Portrait instead of baselines | FG-Portrait · FFHQ cross-reenactment | FID: 113.7 -> 99.4 (vs HunyuanPortrait) | Improved |
Closest papers by meaning
- RealPortrait: Realistic Portrait Animation with Diffusion Transformers (AAAI 2025 · similarity 0.89)
- DiffPortrait3D: Controllable Diffusion for Zero-Shot Portrait View Synthesis (CVPR 2024 · similarity 0.87)
- GuideFlow3D: Optimization-Guided Rectified Flow For Appearance Transfer (NeurIPS 2025 · similarity 0.87)
- 3DToonify: Creating Your High-Fidelity 3D Stylized Avatar Easily from 2D Portrait Images (CVPR 2024 · similarity 0.87)
- High-Fidelity Relightable Monocular Portrait Animation with Lighting-Controllable Video Diffusion Model (CVPR 2025 · similarity 0.87)
- X-NeMo: Expressive Neural Motion Reenactment via Disentangled Latent Attention (ICLR 2025 · 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.
