Research map / Video / Video generation
Lynx: Towards High-Fidelity Personalized Video Generation
Shen Sang, Tiancheng Zhi, Tianpei Gu, Jing Liu, Linjie Luo
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 ID-adapter to Ref-adapter-only model (ablation) | Lynx · Personalized video benchmark (40 subjects x 20 prompts) | ID resemblance (insightface): 0.523 -> 0.699 | Improved |
| add ID-adapter to Ref-adapter-only model (ablation) | Lynx · Personalized video benchmark (40 subjects x 20 prompts) | Prompt following: 0.738 -> 0.722 | Mixed |
| add Ref-adapter to ID-adapter-only model (ablation) | Lynx · Personalized video benchmark (40 subjects x 20 prompts) | ID resemblance (insightface): 0.655 -> 0.699 | Improved |
| add Ref-adapter to ID-adapter-only model (ablation) | Lynx · Personalized video benchmark (40 subjects x 20 prompts) | Prompt following: 0.624 -> 0.722 | Improved |
| add Ref-adapter to ID-adapter-only model (ablation) | Lynx · Personalized video benchmark (40 subjects x 20 prompts) | Video quality: 0.925 -> 0.956 | Improved |
| add ID-adapter to Ref-adapter-only model (ablation) | Lynx · Personalized video benchmark (40 subjects x 20 prompts) | Video quality: 0.921 -> 0.956 | Improved |
| initialize Perceiver Resampler from image-domain checkpoint (InstantID) instead of scratch | Lynx ID-adapter · Personalized video benchmark (40 subjects x 20 prompts) | facial resemblance | Improved |
| Lynx vs SkyReels-A2 on identity | Lynx · Personalized video benchmark (40 subjects x 20 prompts) | Face resemblance (insightface): 0.678 -> 0.699 | Improved |
| Lynx vs VACE on motion naturalness | Lynx · Personalized video benchmark (40 subjects x 20 prompts) | Motion naturalness: 0.851 -> 0.837 | Hurt |
Closest papers by meaning
- FaithfulFaces: Pose-Faithful Facial Identity Preservation for Text-to-Video Generation (NeurIPS 2026 · similarity 0.86)
- Identity-Preserving Text-to-Video Generation by Frequency Decomposition (CVPR 2025 · similarity 0.86)
- Multi-subject Open-set Personalization in Video Generation (CVPR 2025 · similarity 0.86)
- MimicTalk: Mimicking a personalized and expressive 3D talking face in minutes (NeurIPS 2024 · similarity 0.86)
- VLOGGER: Multimodal Diffusion for Embodied Avatar Synthesis (CVPR 2025 · similarity 0.85)
- StableAnimator: High-Quality Identity-Preserving Human Image Animation (CVPR 2025 · similarity 0.85)
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.
