Research map / 3D vision / Human pose estimation
PIAvatar: Physically Interactive Avatars via Deformation Gradient Decoupling
Sang-Hun Han, Min-Gyu Park, Jisu Shin, Seunghyun Shin, JINHWI PARK, Hae-Gon Jeon
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 decoupled kinematic velocity and skeleton tracking instead of basic MPM | PIAvatar · Avatar pose tracking, Animatable Gaussians, 100 steps | MSE (m): 0.079 -> 0.019 | Improved |
| use decoupled kinematic velocity instead of basic MPM | PIAvatar · Avatar pose tracking, Animatable Gaussians, 400 steps | MSE (m): 0.332 -> 0.046 | Improved |
| use decoupled kinematic velocity instead of basic MPM | PIAvatar · Avatar pose tracking, SMPL-X, 100 steps | MSE (m): 0.089 -> 0.013 | Improved |
| use decoupled kinematic velocity instead of basic MPM | PIAvatar · Avatar pose tracking, SMPL-X, 400 steps | MSE (m): 0.248 -> 0.039 | Improved |
| use decoupled kinematic velocity instead of basic MPM | PIAvatar · Avatar pose tracking, Animatable Gaussians, 400 steps | Acc@0.01: 0.020 -> 0.534 | Improved |
| use decoupled kinematic velocity instead of basic MPM | PIAvatar · Avatar pose tracking, SMPL-X, 400 steps | Acc@0.01: 0.036 -> 0.583 | Improved |
| add decoupling and Kabsch tracking overhead to basic MPM | PIAvatar · Runtime, Single SMPL-X | time per step (ms): 0.170 -> 0.492 | Hurt |
| add decoupling and Kabsch tracking overhead to basic MPM | PIAvatar · Runtime, Four AG | time per step (ms): 11.015 -> 15.259 | Hurt |
| add shape-preservation term (ablation) | PIAvatar · Belly-ball contact (qualitative) | residual dent | Improved |
| add multi-field contact instead of single-field MPM contact (ablation) | PIAvatar · Belly-ball contact and leg self-contact (qualitative) | stickiness and penetration | Improved |
| optimize density with VLM-reward Bayesian optimization toward heavy vs light | PIAvatar · Ball mass BO with Qwen2.5-VL | converged mass (kg): 8.8 (heavy) vs 0.9 (light) | Improved |
| optimize Young's modulus with VLM-reward Bayesian optimization toward stiff vs soft | PIAvatar · Belly stiffness BO with Qwen2.5-VL | converged E (Pa): 3.38x10^6 (stiff) vs 3.26x10^2 (soft) | Improved |
Closest papers by meaning
- MPMAvatar: Learning 3D Gaussian Avatars with Accurate and Robust Physics-Based Dynamics (NeurIPS 2025 · similarity 0.90)
- AvatarGO: Zero-shot 4D Human-Object Interaction Generation and Animation (ICLR 2025 · similarity 0.87)
- CtrlAvatar: Controllable Avatars Generation via Disentangled Invertible Networks (AAAI 2025 · similarity 0.86)
- SimAvatar: Simulation-Ready Avatars with Layered Hair and Clothing (CVPR 2025 · similarity 0.86)
- RAM-Avatar: Real-time Photo-Realistic Avatar from Monocular Videos with Full-body Control (CVPR 2024 · similarity 0.86)
- TaoAvatar: Real-Time Lifelike Full-Body Talking Avatars for Augmented Reality via 3D Gaussian Splatting (CVPR 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.
