Research map / Video / Video generation
LIA: Latent Image Animator
Yaohui Wang, Di Yang, François Brémond, Antitza Dantcheva
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 orthogonal motion dictionary D_m (LMD) (ablation) | LIA · VoxCeleb | L1: 0.049 -> 0.041 | Improved |
| add the orthogonal motion dictionary D_m (LMD) (ablation) | LIA · VoxCeleb | LPIPS: 0.165 -> 0.123 | Improved |
| add the orthogonal motion dictionary D_m (LMD) (ablation) | LIA · TaichiHD | L1: 0.062 -> 0.057 | Improved |
| add the orthogonal motion dictionary D_m (LMD) (ablation) | LIA · TED-talks | L1: 0.031 -> 0.028 | Improved |
| increase D_m size from 5 to 20 directions (ablation) | LIA · VoxCeleb | L1: 0.051 -> 0.041 | Improved |
| increase D_m size from 5 to 20 directions (ablation) | LIA · TaichiHD | L1: 0.070 -> 0.057 | Improved |
| increase D_m size from 20 to 40 directions (ablation) | LIA · TaichiHD | L1: 0.057 -> 0.060 | Hurt |
| increase D_m size from 20 to 100 directions (ablation) | LIA · VoxCeleb | L1: 0.041 -> 0.041 | No clear effect |
| use LIA instead of MRAA for cross-video motion transfer | LIA · VoxCeleb | video FID: 0.308 -> 0.161 | Improved |
| use LIA instead of MRAA for cross-dataset transfer | LIA · GermanPublicTV | video FID: 0.454 -> 0.406 | Improved |
| use LIA instead of MRAA w/o bg for reconstruction | LIA · TaichiHD | AKD: 5.626 -> 4.823 | Improved |
| use LIA instead of MRAA w/o bg for reconstruction | LIA · TED-talks | AKD: 3.126 -> 3.141 | Hurt |
Closest papers by meaning
- Single Motion Diffusion (ICLR 2024 · similarity 0.87)
- Efficient Video Diffusion Models via Content-Frame Motion-Latent Decomposition (ICLR 2024 · similarity 0.87)
- VidTwin: Video VAE with Decoupled Structure and Dynamics (CVPR 2025 · similarity 0.87)
- X-NeMo: Expressive Neural Motion Reenactment via Disentangled Latent Attention (ICLR 2025 · similarity 0.87)
- AnimateDiff: Animate Your Personalized Text-to-Image Diffusion Models without Specific Tuning (ICLR 2024 · similarity 0.87)
- MotionDreamer: One-to-Many Motion Synthesis with Localized Generative Masked Transformer (ICLR 2025 · 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.
