Research map / 3D vision / 3D generation
CLIP-Actor: Text-Driven Recommendation and Stylization for Animating Human Meshes
Kim Youwang, Kim Ji-Yeon, Tae-Hyun Oh
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 |
|---|---|---|---|
| use hierarchical CLIP then MPNet retrieval instead of single-stage CLIP (ablation) | motion retrieval · SICK[4.4,4.8] | retrieval precision (%): 81.62 -> 81.90 | Improved |
| use CLIP then MPNet order instead of reversed MPNet then CLIP (ablation) | hierarchical retrieval · SICK4.8 | retrieval precision (%): 91.34 -> 92.24 | Improved |
| add hierarchical re-ranking over single-stage MPNet (ablation) | motion retrieval · SICK4.4 | retrieval precision (%): 83.56 -> 85.75 | Improved |
| add temporal view augmentation (top-3 frames instead of single frame) (ablation) | CLIP-Actor DNSF · text-driven stylization, qualitative | visual quality | Improved |
| add multi-modal content mesh sampling (ablation) | CLIP-Actor DNSF · text-driven stylization, qualitative | visual quality | Improved |
| add mask-weighted embedding attention (ablation) | CLIP-Actor DNSF · text-driven stylization, qualitative | visual quality | Improved |
| add decoupled neural style fields and retrieval-based mesh to Text2Mesh (CLIP-Actor vs base) | CLIP-Actor (base) · user study, 46 users, score 1-5 | user score | Improved |
| provide retrieved motion mesh as initial mesh to Text2Mesh | Text2Mesh · qualitative comparison | visual quality | Improved |
Closest papers by meaning
- CLIP-Actor-X: Text-Driven 4D Human Avatar Generation via Cross-Modal Synthesis-Through-Optimization (TPAMI 2026 · similarity 0.92)
- MotionCLIP: Exposing Human Motion Generation to CLIP Space (ECCV 2022 · similarity 0.87)
- Hierarchical Style-based Networks for Motion Synthesis (ECCV 2020 · similarity 0.87)
- Generative Human Motion Stylization in Latent Space (ICLR 2024 · similarity 0.85)
- SnapMoGen: Human Motion Generation from Expressive Texts (NeurIPS 2025 · similarity 0.85)
- SimMotionEdit: Text-Based Human Motion Editing with Motion Similarity Prediction (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.
