Display Settings
System Controls
Spine & Torso
Right Arm (SKEL Features)
Overview
This interactive viewer demonstrates the core concepts from the paper "From Skin to Skeleton: Towards Biomechanically Accurate 3D Digital Humans". The paper introduces SKEL, a parametric 3D human model that combines a standard deformable skin mesh (like SMPL) with a biomechanically realistic and accurate skeleton.
Traditional models in computer vision and graphics often use simplified skeletons with generic ball-and-socket joints, which don't accurately represent human anatomy. This limits their use in fields like biomechanics or medicine. SKEL addresses this by re-rigging the body mesh with a skeleton that has more realistic degrees of freedom (DoFs). This simulation allows you to explore the difference.
How to Use
- Camera Controls: Click and drag to rotate the view. Use the mouse wheel or a two-finger pinch to zoom. Right-click (or two-finger drag) to pan.
- Display Settings: Toggle the visibility of the skin and skeleton. Use the "Model" button to switch between the advanced SKEL model and a simplified SMPL-like model to see the kinematic differences.
- Pose Sliders: Use the sliders to control different parts of the body. Notice the specific controls for the SKEL model, such as "Scapula" and "Forearm Pronation/Supination", which are not typically separated in simpler models.
- Play Demo: Press this button to cycle through a demonstration of the model's capabilities, highlighting the unique movements possible with SKEL.
- Sound: Toggle the sonification to hear audio feedback as the joints move.
Future Directions
The SKEL model is a significant step towards unifying computer vision and biomechanics. As outlined in the paper, future work could build upon this foundation in several exciting ways:
- Expressive Hands and Face: Extending the biomechanical rigging to include detailed hand and facial skeletons.
- Muscle Integration: Adding muscle geometry and activation to the model, allowing for physics-based simulations and analysis of forces during movement.
- Improved Skin Deformation: Learning more accurate skin deformations (pose-dependent blend shapes) directly from the new biomechanical parameters, rather than inheriting them from SMPL.
- Clinical Validation: Using the model in clinical settings to diagnose injuries or diseases and comparing its accuracy against traditional motion capture methods.