Research map / Video / Video generation
Follow-Your-Emoji-Faster: Towards Efficient, Fine-Controllable, and Expressive Freestyle Portrait Animation
Yue Ma, Zexuan Yan, Hongyu Liu, Hongfa Wang, Heng Pan, Yingqing He, Junkun Yuan, Ailing Zeng, Chengfei Cai, Heung‐Yeung Shum, Zhen Li, Wei Liu and 2 more
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 expression mask to the facial fine-grained loss (ablation) | Follow-Your-Emoji-Faster · EmojiBench++ cross reenactment | landmark accuracy: 27.52 -> 38.26 | Improved |
| add identity (facial) mask to the facial fine-grained loss (ablation) | Follow-Your-Emoji-Faster · EmojiBench++ cross reenactment | ID similarity: 0.557 -> 0.761 | Improved |
| add progressive strategy for long video (ablation) | Follow-Your-Emoji-Faster · EmojiBench++ self reenactment | FVD: 135.2 -> 115.4 | Improved |
| use expression-aware landmarks instead of standard 2D landmarks (ablation) | Follow-Your-Emoji-Faster · EmojiBench++ cross reenactment | landmark accuracy: 15.92 -> 38.26 | Improved |
| exclude facial contour points from landmarks (ablation) | Follow-Your-Emoji-Faster · EmojiBench++ cross reenactment | ID similarity: 0.634 -> 0.761 | Improved |
| add pupil points to landmarks (ablation) | Follow-Your-Emoji-Faster · EmojiBench++ self reenactment | FVD: 122.9 -> 115.4 | Improved |
| add Taylor interpolation to the cache (ablation) | Follow-Your-Emoji-Faster · EmojiBench++ self reenactment | SSIM: 0.782 -> 0.829 | Improved |
| add landmark-mask guidance to the cache (ablation) | Follow-Your-Emoji-Faster · EmojiBench++ cross reenactment | ID similarity: 0.711 -> 0.761 | Improved |
| increase cache refresh interval N from 2 to 3 (ablation) | TIC · EmojiBench++ acceleration | speedup: 1.71x -> 2.58x | Mixed |
| raise Taylor order O from 1 to 3 at N=2 (ablation) | TIC · EmojiBench++ cross reenactment | landmark accuracy: 36.92 -> 37.83 | Improved |
| add landmark mask to TIC with N=3 (ablation) | TIC · EmojiBench++ self reenactment | FVD: 127.3 -> 115.4 | Improved |
| halve the denoising steps to speed up inference | Follow-Your-Emoji · EmojiBench++ self reenactment | FVD: 118.7 -> 188.6 | Hurt |
Closest papers by meaning
- FaceShot: Bring Any Character into Life (ICLR 2025 · similarity 0.87)
- StableAnimator: High-Quality Identity-Preserving Human Image Animation (CVPR 2025 · similarity 0.86)
- RealPortrait: Realistic Portrait Animation with Diffusion Transformers (AAAI 2025 · similarity 0.86)
- X-NeMo: Expressive Neural Motion Reenactment via Disentangled Latent Attention (ICLR 2025 · similarity 0.86)
- MagicAnimate: Temporally Consistent Human Image Animation using Diffusion Model (CVPR 2024 · similarity 0.86)
- AnimateDiff: Animate Your Personalized Text-to-Image Diffusion Models without Specific Tuning (ICLR 2024 · similarity 0.86)
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.
