Skeletal Animation

What is Skeletal Animation

Skeletal animation is an animation technique that uses a hierarchical structure of bones to drive the deformation of a 3D model mesh. Unlike per-vertex animation, skeletal animation binds mesh vertices to a virtual skeleton system and moves vertices indirectly by rotating and translating bones, producing smooth and natural motions. This technique has become the mainstream method for character animation in modern games due to its efficiency and flexibility.

Core Concepts

Bones and Joints

Bones consist of a series of joints, typically arranged in a tree hierarchy. The root bone is the starting point of the animation, and child bones inherit transformations from their parents. Each bone stores a transformation relative to its parent (position, rotation, scale), and a global transformation is obtained by multiplying matrices along the hierarchy.

Skinning

Skinning is the process of associating mesh vertices with bones. Each vertex can be influenced by one or more bones, and influence weights determine the effect of bone movement on the vertex. The final vertex position is a weighted combination of all influencing bone transformations.

Bind Pose

The bind pose is the initial skeleton position, usually T-pose or A-pose. Skinning calculations are based on vertex positions and bone spatial relationships in bind pose.

Skinning Algorithms

Linear Blend Skinning

Most commonly used skinning algorithm. For vertex V influenced by bone i with weight wi, the deformed position is:

`V' = Σ wi * Mi * V`

where Mi is the global transformation matrix of bone i multiplied by the inverse bind pose matrix. Linear blend skinning is simple but may cause "candy wrapper" artifacts (volume collapse) at joint bends.

Dual Quaternion Skinning

Uses dual quaternions to blend transformations, avoiding rotation interpolation issues of linear blending and reducing volume loss, maintaining better volume. Slightly more complex but yields better results.

Spherical Blend Skinning

Blends around rotation centers, further improving joint deformation, but higher computational cost.

Animation Playback and Blending

Skeletal animation data is typically stored as a sequence of keyframes, each recording transformations of bones. Playback interpolates between keyframes (linear, spline, etc.) for smooth motion. Modern engines support animation blending, layering, and state machines to control complex character behaviors.

GPU Implementation

Skeletal animation is usually performed in the vertex shader. The bone transformation matrix array (Bone Matrix Palette) is passed to the shader, and vertices are transformed based on influencing bone indices and weights. Due to GPU parallel processing, performance is high. For many bones or vertices, skinning may become a bottleneck, and compute shaders can be used for optimization.

Advantages and Limitations

**Advantages**:

• Small memory footprint compared to per-vertex animation (only bone transforms, not per-frame vertex positions).
• Animation data can be reused; the same skeleton can drive different meshes.
• Supports advanced controls like animation blending and IK (Inverse Kinematics).
• Suitable for real-time rendering with good performance.

**Limitations**:

• Not good at non-rigid deformations like cloth and fluids; needs physics simulation or mesh deformation.
• Skin weight painting is time-consuming and requires artistic skill.
• Linear blend skinning artifacts in extreme poses.

Related Tools and Pipeline

Common modeling software (Maya, Blender, 3ds Max) support bone creation, skinning, and animation. Game engines (Unity, Unreal) provide pipelines for importing and managing skeletal animation. Procedural animation and motion capture are complementary techniques.