Research map / 3D vision / Face and shape
LUNA: Learning Universal 3D Human Animation Beyond Skinning
Peng Li, Rawal Khirodkar, Junxuan Li, Yuan Dong, Chen Cao, Yuan Liu, Wenhan Luo, Yike Guo, Shunsuke Saito
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 LBS structural distillation regularization (ablation) | LUNA · iPhone1K | PSNR: 21.185 -> 24.136 | Improved |
| add LBS structural distillation regularization (ablation) | LUNA · dome data | PSNR: 21.712 -> 24.374 | Improved |
| add decoupled global rotation (ablation) | LUNA · iPhone1K | PSNR: 23.102 -> 24.136 | Improved |
| add decoupled global rotation (ablation) | LUNA · dome data | PSNR: 23.294 -> 24.374 | Improved |
| add multiview finetuning (ablation) | LUNA · iPhone1K | PSNR: 23.931 -> 24.136 | Improved |
| add multiview finetuning (ablation) | LUNA · dome data | LPIPS: 0.099 -> 0.093 | Improved |
| add LBS structural distillation regularization (ablation) | LUNA · iPhone1K | LPIPS: 0.145 -> 0.094 | Improved |
| replace LBS-based animation with LBS-free neural animator | LUNA vs MV-LHM · Motion quality (MSJ) | MSJ: 0.0144 -> 0.0032 | Improved |
| replace LBS-based animation with neural animator | LUNA vs MV-LHM · Motion quality (MAE) | MAE: 0.0477 -> 0.0225 | Improved |
| use LUNA instead of best baseline on non-rigid garments | LUNA vs MV-LHM · Cloth10K | PSNR: 20.124 -> 22.072 | Improved |
| use LUNA instead of ExAvatar on NeuMan | LUNA vs ExAvatar · NeuMan | PSNR: 31.270 -> 26.819 | Hurt |
| drive with another identity (cross-ID) instead of self | LUNA · NeuMan | PSNR: 26.819 -> 26.800 | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- Vid2Avatar — Vid2Avatar: 3D Avatar Reconstruction From Videos in the Wild via Self-Supervised Scene Decomposition (CVPR 2023) · better on 3 of 4 shared results
- IDOL — IDOL: Inertial Deep Orientation-Estimation and Localization (AAAI 2021) · better on 4 of 4 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Instant Expressive Gaussian Head Avatars at Over 100 FPS (ECCV 2026 · similarity 0.86)
- BANMo: Building Animatable 3D Neural Models From Many Casual Videos (CVPR 2022 · similarity 0.86)
- DreamActor-M1: Holistic, Expressive and Robust Human Image Animation with Hybrid Guidance (ICCV 2025 · similarity 0.86)
- MPMAvatar: Learning 3D Gaussian Avatars with Accurate and Robust Physics-Based Dynamics (NeurIPS 2025 · similarity 0.84)
- TapMo: Shape-aware Motion Generation of Skeleton-free Characters (ICLR 2024 · similarity 0.84)
- RigAnyFace: Scaling Neural Facial Mesh Auto-Rigging with Unlabeled Data (NeurIPS 2025 · similarity 0.84)
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.
